├── .gitattributes ├── .gitignore ├── LICENSE ├── README.md ├── _init_paths.py ├── data ├── __init__.py ├── config.py ├── example.jpg ├── scripts │ ├── VOC2007.sh │ └── VOC2012.sh └── voc0712.py ├── demo ├── __init__.py ├── demo.ipynb └── live.py ├── doc ├── SSD.jpg ├── detection_example.png ├── detection_example2.png ├── detection_examples.png └── ssd.png ├── eval.py ├── eval_rpn.py ├── layers ├── __init__.py ├── box_utils.py ├── functions │ ├── __init__.py │ ├── detection.py │ ├── post_rois.py │ └── prior_box.py └── modules │ ├── __init__.py │ ├── l2norm.py │ └── multibox_loss.py ├── lib ├── datasets │ ├── VOCdevkit-matlab-wrapper │ │ ├── get_voc_opts.m │ │ ├── voc_eval.m │ │ └── xVOCap.m │ ├── __init__.py │ ├── __init__.pyc │ ├── __pycache__ │ │ └── __init__.cpython-36.pyc │ ├── coco.py │ ├── coco.pyc │ ├── ds_utils.py │ ├── ds_utils.pyc │ ├── factory.py │ ├── factory.pyc │ ├── imagenet.py │ ├── imagenet.pyc │ ├── imdb.py │ ├── imdb.pyc │ ├── pascal_voc.py │ ├── pascal_voc.pyc │ ├── pascal_voc_rbg.py │ ├── tools │ │ └── mcg_munge.py │ ├── vg.py │ ├── vg.pyc │ ├── vg_eval.py │ ├── vg_eval.pyc │ ├── voc_eval.py │ └── voc_eval.pyc ├── make.sh ├── model │ ├── __init__.py │ ├── __init__.pyc │ ├── __pycache__ │ │ └── __init__.cpython-36.pyc │ ├── faster_rcnn │ │ ├── __init__.py │ │ ├── __init__.pyc │ │ ├── faster_rcnn.py │ │ ├── faster_rcnn.pyc │ │ ├── resnet.py │ │ ├── resnet.pyc │ │ ├── vgg16.py │ │ └── vgg16.pyc │ ├── fixed_proj.py │ ├── fixed_proj.pyc │ ├── nms │ │ ├── .gitignore │ │ ├── __init__.py │ │ ├── __init__.pyc │ │ ├── _ext │ │ │ ├── __init__.py │ │ │ ├── __init__.pyc │ │ │ └── nms │ │ │ │ ├── __init__.py │ │ │ │ ├── __init__.pyc │ │ │ │ └── _nms.so │ │ ├── build.py │ │ ├── make.sh │ │ ├── nms_gpu.py │ │ ├── nms_gpu.pyc │ │ ├── nms_kernel.cu │ │ ├── nms_wrapper.py │ │ ├── nms_wrapper.pyc │ │ └── src │ │ │ ├── nms_cuda.c │ │ │ ├── nms_cuda.h │ │ │ ├── nms_cuda_kernel.cu │ │ │ ├── nms_cuda_kernel.cu.o │ │ │ └── nms_cuda_kernel.h │ ├── roi_align │ │ ├── __init__.py │ │ ├── __init__.pyc │ │ ├── _ext │ │ │ ├── __init__.py │ │ │ ├── __init__.pyc │ │ │ └── roi_align │ │ │ │ ├── __init__.py │ │ │ │ ├── __init__.pyc │ │ │ │ └── _roi_align.so │ │ ├── build.py │ │ ├── functions │ │ │ ├── __init__.py │ │ │ ├── __init__.pyc │ │ │ ├── roi_align.py │ │ │ └── roi_align.pyc │ │ ├── make.sh │ │ ├── modules │ │ │ ├── __init__.py │ │ │ ├── __init__.pyc │ │ │ ├── roi_align.py │ │ │ └── roi_align.pyc │ │ └── src │ │ │ ├── roi_align_cuda.c │ │ │ ├── roi_align_cuda.h │ │ │ ├── roi_align_kernel.cu │ │ │ ├── roi_align_kernel.cu.o │ │ │ └── roi_align_kernel.h │ ├── roi_crop │ │ ├── __init__.py │ │ ├── __init__.pyc │ │ ├── _ext │ │ │ ├── __init__.py │ │ │ ├── __init__.pyc │ │ │ ├── crop_resize │ │ │ │ ├── __init__.py │ │ │ │ └── _crop_resize.so │ │ │ └── roi_crop │ │ │ │ ├── __init__.py │ │ │ │ ├── __init__.pyc │ │ │ │ └── _roi_crop.so │ │ ├── build.py │ │ ├── functions │ │ │ ├── __init__.py │ │ │ ├── __init__.pyc │ │ │ ├── crop_resize.py │ │ │ ├── gridgen.py │ │ │ ├── roi_crop.py │ │ │ └── roi_crop.pyc │ │ ├── make.sh │ │ ├── modules │ │ │ ├── __init__.py │ │ │ ├── __init__.pyc │ │ │ ├── gridgen.py │ │ │ ├── roi_crop.py │ │ │ └── roi_crop.pyc │ │ └── src │ │ │ ├── roi_crop.c │ │ │ ├── roi_crop.h │ │ │ ├── roi_crop_cuda.c │ │ │ ├── roi_crop_cuda.h │ │ │ ├── roi_crop_cuda_kernel.cu │ │ │ ├── roi_crop_cuda_kernel.cu.o │ │ │ └── roi_crop_cuda_kernel.h │ ├── roi_pooling │ │ ├── __init__.py │ │ ├── __init__.pyc │ │ ├── __pycache__ │ │ │ └── __init__.cpython-36.pyc │ │ ├── _ext │ │ │ ├── __init__.py │ │ │ ├── __init__.pyc │ │ │ ├── __pycache__ │ │ │ │ └── __init__.cpython-36.pyc │ │ │ └── roi_pooling │ │ │ │ ├── __init__.py │ │ │ │ ├── __init__.pyc │ │ │ │ ├── __pycache__ │ │ │ │ └── __init__.cpython-36.pyc │ │ │ │ └── _roi_pooling.so │ │ ├── build.py │ │ ├── functions │ │ │ ├── __init__.py │ │ │ ├── __init__.pyc │ │ │ ├── __pycache__ │ │ │ │ ├── __init__.cpython-36.pyc │ │ │ │ └── roi_pool.cpython-36.pyc │ │ │ ├── roi_pool.py │ │ │ └── roi_pool.pyc │ │ ├── modules │ │ │ ├── __init__.py │ │ │ ├── __init__.pyc │ │ │ ├── __pycache__ │ │ │ │ ├── __init__.cpython-36.pyc │ │ │ │ └── roi_pool.cpython-36.pyc │ │ │ ├── roi_pool.py │ │ │ └── roi_pool.pyc │ │ └── src │ │ │ ├── roi_pooling.c │ │ │ ├── roi_pooling.cu.o │ │ │ ├── roi_pooling.h │ │ │ ├── roi_pooling_cuda.c │ │ │ ├── roi_pooling_cuda.h │ │ │ ├── roi_pooling_kernel.cu │ │ │ └── roi_pooling_kernel.h │ ├── roi_pooling_replace.py │ ├── roi_pooling_replace.pyc │ ├── rpn │ │ ├── __init__.py │ │ ├── __init__.pyc │ │ ├── anchor_target_layer.py │ │ ├── anchor_target_layer.pyc │ │ ├── bbox_transform.py │ │ ├── bbox_transform.pyc │ │ ├── generate_anchors.py │ │ ├── generate_anchors.pyc │ │ ├── proposal_layer.py │ │ ├── proposal_layer.pyc │ │ ├── proposal_target_layer_cascade.py │ │ ├── proposal_target_layer_cascade.pyc │ │ ├── rpn.py │ │ └── rpn.pyc │ └── utils │ │ ├── .gitignore │ │ ├── __init__.py │ │ ├── __init__.pyc │ │ ├── __pycache__ │ │ ├── __init__.cpython-36.pyc │ │ └── config.cpython-36.pyc │ │ ├── bbox.c │ │ ├── bbox.pyx │ │ ├── blob.py │ │ ├── blob.pyc │ │ ├── config.py │ │ ├── config.pyc │ │ ├── cython_bbox.cpython-36m-x86_64-linux-gnu.so │ │ ├── cython_bbox.so │ │ ├── logger.py │ │ ├── net_utils.py │ │ └── net_utils.pyc ├── pycocotools │ ├── UPSTREAM_REV │ ├── __init__.py │ ├── __init__.pyc │ ├── _mask.c │ ├── _mask.cpython-36m-x86_64-linux-gnu.so │ ├── _mask.pyx │ ├── _mask.so │ ├── coco.py │ ├── coco.pyc │ ├── cocoeval.py │ ├── cocoeval.pyc │ ├── license.txt │ ├── mask.py │ ├── mask.pyc │ ├── maskApi.c │ └── maskApi.h ├── roi_data_layer │ ├── __init__.py │ ├── __init__.pyc │ ├── __pycache__ │ │ ├── __init__.cpython-36.pyc │ │ └── roidb.cpython-36.pyc │ ├── minibatch.py │ ├── minibatch.pyc │ ├── roibatchLoader.py │ ├── roibatchLoader.pyc │ ├── roidb.py │ └── roidb.pyc └── setup.py ├── lstd_extra_test.py ├── ssd.py ├── test.py ├── test_overlap.py ├── train.py └── utils ├── __init__.py └── augmentations.py /.gitattributes: -------------------------------------------------------------------------------- 1 | *.ipynb linguist-language=Python 2 | .ipynb_checkpoints/* linguist-documentation 3 | dev.ipynb linguist-documentation 4 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *$py.class 5 | 6 | # C extensions 7 | *.so 8 | 9 | # Distribution / packaging 10 | .Python 11 | env/ 12 | build/ 13 | develop-eggs/ 14 | dist/ 15 | downloads/ 16 | eggs/ 17 | .eggs/ 18 | lib/ 19 | lib64/ 20 | parts/ 21 | sdist/ 22 | var/ 23 | *.egg-info/ 24 | .installed.cfg 25 | *.egg 26 | 27 | # PyInstaller 28 | # Usually these files are written by a python script from a template 29 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 30 | *.manifest 31 | *.spec 32 | 33 | # Installer logs 34 | pip-log.txt 35 | pip-delete-this-directory.txt 36 | 37 | # Unit test / coverage reports 38 | htmlcov/ 39 | .tox/ 40 | .coverage 41 | .coverage.* 42 | .cache 43 | nosetests.xml 44 | coverage.xml 45 | *,cover 46 | .hypothesis/ 47 | 48 | # Translations 49 | *.mo 50 | *.pot 51 | 52 | # Django stuff: 53 | *.log 54 | local_settings.py 55 | 56 | # Flask stuff: 57 | instance/ 58 | .webassets-cache 59 | 60 | # Scrapy stuff: 61 | .scrapy 62 | 63 | # Sphinx documentation 64 | docs/_build/ 65 | 66 | # PyBuilder 67 | target/ 68 | 69 | # IPython Notebook 70 | .ipynb_checkpoints 71 | 72 | # pyenv 73 | .python-version 74 | 75 | # celery beat schedule file 76 | celerybeat-schedule 77 | 78 | # dotenv 79 | .env 80 | 81 | # virtualenv 82 | venv/ 83 | ENV/ 84 | 85 | # Spyder project settings 86 | .spyderproject 87 | 88 | # Rope project settings 89 | .ropeproject 90 | 91 | # atom remote-sync package 92 | .remote-sync.json 93 | 94 | # weights 95 | weights/ 96 | 97 | #DS_Store 98 | .DS_Store 99 | 100 | # dev stuff 101 | eval/ 102 | eval.ipynb 103 | dev.ipynb 104 | .vscode/ 105 | 106 | # not ready 107 | videos/ 108 | templates/ 109 | data/ssd_dataloader.py 110 | data/datasets/ 111 | doc/visualize.py 112 | read_results.py 113 | ssd300_120000/ 114 | demos/live 115 | webdemo.py 116 | test_data_aug.py 117 | 118 | # attributes 119 | 120 | # pycharm 121 | .idea/ 122 | 123 | # temp checkout soln 124 | data/datasets/ 125 | data/ssd_dataloader.py 126 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | MIT License 2 | 3 | Copyright (c) 2017 Max deGroot, Ellis Brown 4 | 5 | Permission is hereby granted, free of charge, to any person obtaining a copy 6 | of this software and associated documentation files (the "Software"), to deal 7 | in the Software without restriction, including without limitation the rights 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 9 | copies of the Software, and to permit persons to whom the Software is 10 | furnished to do so, subject to the following conditions: 11 | 12 | The above copyright notice and this permission notice shall be included in all 13 | copies or substantial portions of the Software. 14 | 15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 21 | SOFTWARE. 22 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # pytorch-lstd 2 | -------------------------------------------------------------------------------- /_init_paths.py: -------------------------------------------------------------------------------- 1 | import os.path as osp 2 | import sys 3 | 4 | def add_path(path): 5 | if path not in sys.path: 6 | sys.path.insert(0, path) 7 | 8 | this_dir = osp.dirname(__file__) 9 | 10 | # Add lib to PYTHONPATH 11 | lib_path = osp.join(this_dir, 'lib') 12 | add_path(lib_path) 13 | 14 | -------------------------------------------------------------------------------- /data/__init__.py: -------------------------------------------------------------------------------- 1 | from .voc0712 import VOCDetection, AnnotationTransform, detection_collate, VOC_CLASSES 2 | from .config import * 3 | import cv2 4 | import numpy as np 5 | 6 | 7 | def base_transform(image, size, mean): 8 | x = cv2.resize(image, (size, size)).astype(np.float32) 9 | # x = cv2.resize(np.array(image), (size, size)).astype(np.float32) 10 | x -= mean 11 | x = x.astype(np.float32) 12 | return x 13 | 14 | 15 | class BaseTransform: 16 | def __init__(self, size, mean): 17 | self.size = size 18 | self.mean = np.array(mean, dtype=np.float32) 19 | 20 | def __call__(self, image, boxes=None, labels=None): 21 | return base_transform(image, self.size, self.mean), boxes, labels 22 | -------------------------------------------------------------------------------- /data/config.py: -------------------------------------------------------------------------------- 1 | # config.py 2 | import os.path 3 | 4 | # gets home dir cross platform 5 | home = os.path.expanduser("~") 6 | ddir = os.path.join(home,"data/VOCdevkit/") 7 | 8 | # note: if you used our download scripts, this should be right 9 | VOCroot = ddir # path to VOCdevkit root dir 10 | 11 | # default batch size 12 | BATCHES = 32 13 | # data reshuffled at every epoch 14 | SHUFFLE = True 15 | # number of subprocesses to use for data loading 16 | WORKERS = 4 17 | 18 | 19 | #SSD300 CONFIGS 20 | # newer version: use additional conv11_2 layer as last layer before multibox layers 21 | v2 = { 22 | 'feature_maps' : [38, 19, 10, 5, 3, 1], 23 | 24 | 'min_dim' : 300, 25 | 26 | 'steps' : [8, 16, 32, 64, 100, 300], 27 | 28 | 'min_sizes' : [30, 60, 111, 162, 213, 264], 29 | 30 | 'max_sizes' : [60, 111, 162, 213, 264, 315], 31 | 32 | # 'aspect_ratios' : [[2, 1/2], [2, 1/2, 3, 1/3], [2, 1/2, 3, 1/3], 33 | # [2, 1/2, 3, 1/3], [2, 1/2], [2, 1/2]], 34 | 'aspect_ratios' : [[2], [2, 3], [2, 3], [2, 3], [2], [2]], 35 | 36 | 'variance' : [0.1, 0.2], 37 | 38 | 'clip' : True, 39 | 40 | 'name' : 'v2', 41 | } 42 | 43 | # use average pooling layer as last layer before multibox layers 44 | v1 = { 45 | 'feature_maps' : [38, 19, 10, 5, 3, 1], 46 | 47 | 'min_dim' : 300, 48 | 49 | 'steps' : [8, 16, 32, 64, 100, 300], 50 | 51 | 'min_sizes' : [30, 60, 114, 168, 222, 276], 52 | 53 | 'max_sizes' : [-1, 114, 168, 222, 276, 330], 54 | 55 | # 'aspect_ratios' : [[2], [2, 3], [2, 3], [2, 3], [2, 3], [2, 3]], 56 | 'aspect_ratios' : [[1,1,2,1/2],[1,1,2,1/2,3,1/3],[1,1,2,1/2,3,1/3], 57 | [1,1,2,1/2,3,1/3],[1,1,2,1/2,3,1/3],[1,1,2,1/2,3,1/3]], 58 | 59 | 'variance' : [0.1, 0.2], 60 | 61 | 'clip' : True, 62 | 63 | 'name' : 'v1', 64 | } 65 | -------------------------------------------------------------------------------- /data/example.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/data/example.jpg -------------------------------------------------------------------------------- /data/scripts/VOC2007.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # Ellis Brown 3 | 4 | start=`date +%s` 5 | 6 | # handle optional download dir 7 | if [ -z "$1" ] 8 | then 9 | # navigate to ~/data 10 | echo "navigating to ~/data/ ..." 11 | mkdir -p ~/data 12 | cd ~/data/ 13 | else 14 | # check if is valid directory 15 | if [ ! -d $1 ]; then 16 | echo $1 "is not a valid directory" 17 | exit 0 18 | fi 19 | echo "navigating to" $1 "..." 20 | cd $1 21 | fi 22 | 23 | echo "Downloading VOC2007 trainval ..." 24 | # Download the data. 25 | curl -LO http://host.robots.ox.ac.uk/pascal/VOC/voc2007/VOCtrainval_06-Nov-2007.tar 26 | echo "Downloading VOC2007 test data ..." 27 | curl -LO http://host.robots.ox.ac.uk/pascal/VOC/voc2007/VOCtest_06-Nov-2007.tar 28 | echo "Done downloading." 29 | 30 | # Extract data 31 | echo "Extracting trainval ..." 32 | tar -xvf VOCtrainval_06-Nov-2007.tar 33 | echo "Extracting test ..." 34 | tar -xvf VOCtest_06-Nov-2007.tar 35 | echo "removing tars ..." 36 | rm VOCtrainval_06-Nov-2007.tar 37 | rm VOCtest_06-Nov-2007.tar 38 | 39 | end=`date +%s` 40 | runtime=$((end-start)) 41 | 42 | echo "Completed in" $runtime "seconds" -------------------------------------------------------------------------------- /data/scripts/VOC2012.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # Ellis Brown 3 | 4 | start=`date +%s` 5 | 6 | # handle optional download dir 7 | if [ -z "$1" ] 8 | then 9 | # navigate to ~/data 10 | echo "navigating to ~/data/ ..." 11 | mkdir -p ~/data 12 | cd ~/data/ 13 | else 14 | # check if is valid directory 15 | if [ ! -d $1 ]; then 16 | echo $1 "is not a valid directory" 17 | exit 0 18 | fi 19 | echo "navigating to" $1 "..." 20 | cd $1 21 | fi 22 | 23 | echo "Downloading VOC2012 trainval ..." 24 | # Download the data. 25 | curl -LO http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar 26 | echo "Done downloading." 27 | 28 | 29 | # Extract data 30 | echo "Extracting trainval ..." 31 | tar -xvf VOCtrainval_11-May-2012.tar 32 | echo "removing tar ..." 33 | rm VOCtrainval_11-May-2012.tar 34 | 35 | end=`date +%s` 36 | runtime=$((end-start)) 37 | 38 | echo "Completed in" $runtime "seconds" -------------------------------------------------------------------------------- /demo/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/demo/__init__.py -------------------------------------------------------------------------------- /demo/live.py: -------------------------------------------------------------------------------- 1 | from __future__ import print_function 2 | import torch 3 | from torch.autograd import Variable 4 | import cv2 5 | import time 6 | from imutils.video import FPS, WebcamVideoStream 7 | import argparse 8 | 9 | parser = argparse.ArgumentParser(description='Single Shot MultiBox Detection') 10 | parser.add_argument('--weights', default='weights/ssd_300_VOC0712.pth', 11 | type=str, help='Trained state_dict file path') 12 | parser.add_argument('--cuda', default=False, type=bool, 13 | help='Use cuda to train model') 14 | args = parser.parse_args() 15 | 16 | COLORS = [(255, 0, 0), (0, 255, 0), (0, 0, 255)] 17 | FONT = cv2.FONT_HERSHEY_SIMPLEX 18 | 19 | 20 | def cv2_demo(net, transform): 21 | def predict(frame): 22 | height, width = frame.shape[:2] 23 | x = torch.from_numpy(transform(frame)[0]).permute(2, 0, 1) 24 | x = Variable(x.unsqueeze(0)) 25 | y = net(x) # forward pass 26 | detections = y.data 27 | # scale each detection back up to the image 28 | scale = torch.Tensor([width, height, width, height]) 29 | for i in range(detections.size(1)): 30 | j = 0 31 | while detections[0, i, j, 0] >= 0.6: 32 | pt = (detections[0, i, j, 1:] * scale).cpu().numpy() 33 | cv2.rectangle(frame, (int(pt[0]), int(pt[1])), (int(pt[2]), 34 | int(pt[3])), COLORS[i % 3], 2) 35 | cv2.putText(frame, labelmap[i - 1], (int(pt[0]), int(pt[1])), FONT, 36 | 2, (255, 255, 255), 2, cv2.LINE_AA) 37 | j += 1 38 | return frame 39 | 40 | # start video stream thread, allow buffer to fill 41 | print("[INFO] starting threaded video stream...") 42 | stream = WebcamVideoStream(src=0).start() # default camera 43 | time.sleep(1.0) 44 | # start fps timer 45 | # loop over frames from the video file stream 46 | while True: 47 | # grab next frame 48 | frame = stream.read() 49 | key = cv2.waitKey(1) & 0xFF 50 | 51 | # update FPS counter 52 | fps.update() 53 | frame = predict(frame) 54 | 55 | # keybindings for display 56 | if key == ord('p'): # pause 57 | while True: 58 | key2 = cv2.waitKey(1) or 0xff 59 | cv2.imshow('frame', frame) 60 | if key2 == ord('p'): # resume 61 | break 62 | cv2.imshow('frame', frame) 63 | if key == 27: # exit 64 | break 65 | 66 | 67 | if __name__ == '__main__': 68 | import sys 69 | from os import path 70 | sys.path.append(path.dirname(path.dirname(path.abspath(__file__)))) 71 | 72 | from data import BaseTransform, VOC_CLASSES as labelmap 73 | from ssd import build_ssd 74 | 75 | net = build_ssd('test', 300, 21) # initialize SSD 76 | net.load_state_dict(torch.load(args.weights)) 77 | transform = BaseTransform(net.size, (104/256.0, 117/256.0, 123/256.0)) 78 | 79 | fps = FPS().start() 80 | # stop the timer and display FPS information 81 | cv2_demo(net.eval(), transform) 82 | fps.stop() 83 | 84 | print("[INFO] elasped time: {:.2f}".format(fps.elapsed())) 85 | print("[INFO] approx. FPS: {:.2f}".format(fps.fps())) 86 | 87 | # cleanup 88 | cv2.destroyAllWindows() 89 | stream.stop() 90 | -------------------------------------------------------------------------------- /doc/SSD.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/doc/SSD.jpg -------------------------------------------------------------------------------- /doc/detection_example.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/doc/detection_example.png -------------------------------------------------------------------------------- /doc/detection_example2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/doc/detection_example2.png -------------------------------------------------------------------------------- /doc/detection_examples.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/doc/detection_examples.png -------------------------------------------------------------------------------- /doc/ssd.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/doc/ssd.png -------------------------------------------------------------------------------- /layers/__init__.py: -------------------------------------------------------------------------------- 1 | from .functions import * 2 | from .modules import * 3 | -------------------------------------------------------------------------------- /layers/functions/__init__.py: -------------------------------------------------------------------------------- 1 | from .detection import Detect 2 | from .prior_box import PriorBox 3 | from .post_rois import Post_rois 4 | 5 | __all__ = ['Detect', 'PriorBox', 'Post_rois'] 6 | -------------------------------------------------------------------------------- /layers/functions/detection.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch.autograd import Function 3 | from ..box_utils import decode, nms 4 | from data import v2 as cfg 5 | 6 | 7 | class Detect(Function): 8 | """At test time, Detect is the final layer of SSD. Decode location preds, 9 | apply non-maximum suppression to location predictions based on conf 10 | scores and threshold to a top_k number of output predictions for both 11 | confidence score and locations. 12 | """ 13 | def __init__(self, num_classes, bkg_label, top_k, conf_thresh, nms_thresh): 14 | self.num_classes = num_classes 15 | self.background_label = bkg_label 16 | self.top_k = top_k 17 | # Parameters used in nms. 18 | self.nms_thresh = nms_thresh 19 | if nms_thresh <= 0: 20 | raise ValueError('nms_threshold must be non negative.') 21 | self.conf_thresh = conf_thresh 22 | self.variance = cfg['variance'] 23 | 24 | def forward(self, loc_data, conf_data, prior_data): 25 | """ 26 | Args: 27 | loc_data: (tensor) Loc preds from loc layers 28 | Shape: [batch,num_priors*4] 29 | conf_data: (tensor) Shape: Conf preds from conf layers 30 | Shape: [batch*num_priors,num_classes] 31 | prior_data: (tensor) Prior boxes and variances from priorbox layers 32 | Shape: [1,num_priors,4] 33 | """ 34 | num = loc_data.size(0) # batch size 35 | num_priors = prior_data.size(0) 36 | output = torch.zeros(num, self.num_classes, self.top_k, 5) 37 | conf_preds = conf_data.view(num, num_priors, 38 | self.num_classes).transpose(2, 1) 39 | 40 | # Decode predictions into bboxes. 41 | for i in range(num): 42 | decoded_boxes = decode(loc_data[i], prior_data, self.variance) 43 | # For each class, perform nms 44 | conf_scores = conf_preds[i].clone() 45 | 46 | for cl in range(1, self.num_classes): 47 | c_mask = conf_scores[cl].gt(self.conf_thresh) 48 | scores = conf_scores[cl][c_mask] 49 | if scores.dim() == 0: 50 | continue 51 | l_mask = c_mask.unsqueeze(1).expand_as(decoded_boxes) 52 | boxes = decoded_boxes[l_mask].view(-1, 4) 53 | # idx of highest scoring and non-overlapping boxes per class 54 | ids, count = nms(boxes, scores, self.nms_thresh, self.top_k) 55 | output[i, cl, :count] = \ 56 | torch.cat((scores[ids[:count]].unsqueeze(1), 57 | boxes[ids[:count]]), 1) 58 | flt = output.contiguous().view(num, -1, 5) 59 | _, idx = flt[:, :, 0].sort(1, descending=True) 60 | _, rank = idx.sort(1) 61 | flt[(rank < self.top_k).unsqueeze(-1).expand_as(flt)].fill_(0) 62 | return output 63 | -------------------------------------------------------------------------------- /layers/functions/post_rois.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch.autograd import Function 3 | from ..box_utils import decode, nms 4 | from data import v2 as cfg 5 | import cv2 6 | 7 | class Post_rois(Function): 8 | """At test time, Detect is the final layer of SSD. Decode location preds, 9 | apply non-maximum suppression to location predictions based on conf 10 | scores and threshold to a top_k number of output predictions for both 11 | confidence score and locations. 12 | """ 13 | def __init__(self, num_classes, bkg_label, top_k, conf_thresh, nms_thresh): 14 | self.num_classes = num_classes 15 | self.background_label = bkg_label 16 | self.top_k = top_k 17 | # Parameters used in nms. 18 | self.nms_thresh = nms_thresh 19 | if nms_thresh <= 0: 20 | raise ValueError('nms_threshold must be non negative.') 21 | self.conf_thresh = conf_thresh 22 | self.variance = cfg['variance'] 23 | 24 | def forward(self, img, loc_data, conf_data, prior_data): 25 | """ 26 | Args: 27 | loc_data: (tensor) Loc preds from loc layers 28 | Shape: [batch,num_priors*4] 29 | conf_data: (tensor) Shape: Conf preds from conf layers 30 | Shape: [batch*num_priors,num_classes] 31 | prior_data: (tensor) Prior boxes and variances from priorbox layers 32 | Shape: [1,num_priors,4] 33 | """ 34 | num = loc_data.size(0) # batch size 35 | num_priors = prior_data.size(0) # 8732 36 | output = torch.zeros(num, self.num_classes-1, self.top_k, 5) 37 | conf_preds = conf_data.view(num, num_priors, 38 | self.num_classes).transpose(2, 1) 39 | 40 | #print(num) 41 | #print(prior_data.size()) 8732, 4 42 | #print(conf_data.size()) 8732, 2 43 | #print(loc_data.size()) 44 | 45 | def vis(img, prior, decoded_boxes, confs, idx): 46 | im = img.cpu() 47 | im = im.numpy() 48 | im = im.transpose(1,2,0) 49 | img = im[:, :, (2, 1, 0)] 50 | im = img.copy() 51 | write_flag = False 52 | ''' 53 | for j in range(4): 54 | write_flag = True 55 | x_c = int(prior[j][0]*300) 56 | y_c = int(prior[j][1]*300) 57 | h = int(prior[j][2]*300) 58 | w = int(prior[j][3]*300) 59 | cv2.rectangle(im, (x_c - int(w/2), y_c - int(h/2)), (x_c + int(w/2), y_c + int(h/2)),(255,255,255),1) 60 | ''' 61 | #for i in range(decoded_boxes.size(0)): 62 | for i in range(10): 63 | write_flag = True 64 | cv2.rectangle(im, (int(decoded_boxes[i][0]*300), int(decoded_boxes[i][1]*300)),(int(decoded_boxes[i][2]*300),int(decoded_boxes[i][3]*300)), (255,0,0),1) 65 | cv2.putText(im, str(confs[i]), (int(decoded_boxes[i][0]*300) + int((decoded_boxes[i][2]*300 - decoded_boxes[i][0]*300)/2) ,int(decoded_boxes[i][1]*300) + int((decoded_boxes[i][3]*300 - decoded_boxes[i][1]*300)/2)), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0,0,255)) 66 | ''' 67 | for j in range(proposal.size(0)): 68 | if cls_t[j]>0: 69 | write_flag = True 70 | cv2.rectangle(im, (int(proposal[j][0]*300), int(proposal[j][1]*300)), (int(proposal[j][2]*300), int(proposal[j][3]*300)),(255,0,0),2) 71 | ''' 72 | if write_flag: 73 | cv2.imwrite('./vis/'+str(idx)+'.jpg', im) 74 | #cv2.imshow('./vis/'+str(idx)+'.jpg', im) 75 | 76 | # Decode predictions into bboxes. 77 | for i in range(num): 78 | assert prior_data.cpu().numpy().all() >= 0 79 | prior_data = prior_data.cuda(loc_data[i].get_device()) 80 | decoded_boxes = decode(loc_data[i], prior_data, self.variance) 81 | 82 | # For each class, perform nms 83 | conf_scores = conf_preds[i][1].clone() 84 | # filter 85 | 86 | # apply nms 87 | ids, count = nms(decoded_boxes, conf_scores, self.nms_thresh, 1000) 88 | 89 | # sort all conf_scores from high to low 90 | sort_score, sort_index = torch.sort(conf_scores[ids[:count]], descending=True) 91 | # get top 100 92 | sort_index = sort_index[0:100] 93 | scores = conf_scores[ids[:count]][sort_index] 94 | decoded_boxes = decoded_boxes[ids[:count]][sort_index,:] 95 | confs = scores.clone() 96 | # change score to img index 97 | scores[:] = i 98 | output[i, 0, :100]= torch.cat((scores.unsqueeze(1), decoded_boxes), 1) 99 | #output[i, 0, :count]= torch.cat((scores[ids[:count]].unsqueeze(1), decoded_boxes[ids[:count]]), 1) 100 | ''' 101 | for cl in range(1, self.num_classes): 102 | c_mask = conf_scores[cl].gt(self.conf_thresh) 103 | c_mask[:] = 0 104 | # sort all conf_scores from highest to lowest 105 | sort_score, sort_index = torch.sort(conf_scores[cl], descending=True) 106 | sort_index = sort_index[0:1000] 107 | c_mask[sort_index] = 1 108 | # get top 1000 boxes 109 | scores = conf_scores[cl][c_mask] 110 | if scores.dim() == 0: 111 | continue 112 | 113 | l_mask = c_mask.unsqueeze(1).expand_as(decoded_boxes) 114 | boxes = decoded_boxes[l_mask].view(-1, 4) 115 | 116 | #clip 117 | #boxes = torch.clamp(boxes, 0, 1) 118 | 119 | # idx of highest scoring and non-overlapping boxes per class 120 | ids, count = nms(boxes, scores, self.nms_thresh, self.top_k) 121 | scores[:] = i 122 | output[i, (cl-1) , :count] = \ 123 | torch.cat((scores[ids[:count]].unsqueeze(1), 124 | boxes[ids[:count]]), 1) 125 | 126 | ''' 127 | #truths = targets[i][:, :-1].data 128 | #labels = targets[i][:, -1].data 129 | 130 | if 0: 131 | vis(img[i], prior_data, decoded_boxes, confs, str(i)) 132 | 133 | 134 | return output 135 | 136 | -------------------------------------------------------------------------------- /layers/functions/prior_box.py: -------------------------------------------------------------------------------- 1 | from __future__ import division 2 | import torch 3 | from math import sqrt as sqrt 4 | from itertools import product as product 5 | 6 | class PriorBox(object): 7 | """Compute priorbox coordinates in center-offset form for each source 8 | feature map. 9 | Note: 10 | This 'layer' has changed between versions of the original SSD 11 | paper, so we include both versions, but note v2 is the most tested and most 12 | recent version of the paper. 13 | 14 | """ 15 | def __init__(self, cfg): 16 | super(PriorBox, self).__init__() 17 | # self.type = cfg.name 18 | self.image_size = cfg['min_dim'] 19 | # number of priors for feature map location (either 4 or 6) 20 | self.num_priors = len(cfg['aspect_ratios']) 21 | self.variance = cfg['variance'] or [0.1] 22 | self.feature_maps = cfg['feature_maps'] 23 | self.min_sizes = cfg['min_sizes'] 24 | self.max_sizes = cfg['max_sizes'] 25 | self.steps = cfg['steps'] 26 | self.aspect_ratios = cfg['aspect_ratios'] 27 | self.clip = cfg['clip'] 28 | self.version = cfg['name'] 29 | for v in self.variance: 30 | if v <= 0: 31 | raise ValueError('Variances must be greater than 0') 32 | 33 | def forward(self): 34 | mean = [] 35 | # TODO merge these 36 | if self.version == 'v2': 37 | for k, f in enumerate(self.feature_maps): 38 | for i, j in product(range(f), repeat=2): 39 | f_k = self.image_size / self.steps[k] 40 | # unit center x,y 41 | cx = (j + 0.5) / f_k 42 | cy = (i + 0.5) / f_k 43 | 44 | # aspect_ratio: 1 45 | # rel size: min_size 46 | s_k = self.min_sizes[k]/self.image_size 47 | mean += [cx, cy, s_k, s_k] 48 | 49 | # aspect_ratio: 1 50 | # rel size: sqrt(s_k * s_(k+1)) 51 | s_k_prime = sqrt(s_k * (self.max_sizes[k]/self.image_size)) 52 | mean += [cx, cy, s_k_prime, s_k_prime] 53 | 54 | # rest of aspect ratios 55 | for ar in self.aspect_ratios[k]: 56 | mean += [cx, cy, s_k*sqrt(ar), s_k/sqrt(ar)] 57 | mean += [cx, cy, s_k/sqrt(ar), s_k*sqrt(ar)] 58 | 59 | else: 60 | # original version generation of prior (default) boxes 61 | for i, k in enumerate(self.feature_maps): 62 | step_x = step_y = self.image_size/k 63 | for h, w in product(range(k), repeat=2): 64 | c_x = ((w+0.5) * step_x) 65 | c_y = ((h+0.5) * step_y) 66 | c_w = c_h = self.min_sizes[i] / 2 67 | s_k = self.image_size # 300 68 | # aspect_ratio: 1, 69 | # size: min_size 70 | mean += [(c_x-c_w)/s_k, (c_y-c_h)/s_k, 71 | (c_x+c_w)/s_k, (c_y+c_h)/s_k] 72 | if self.max_sizes[i] > 0: 73 | # aspect_ratio: 1 74 | # size: sqrt(min_size * max_size)/2 75 | c_w = c_h = sqrt(self.min_sizes[i] * 76 | self.max_sizes[i])/2 77 | mean += [(c_x-c_w)/s_k, (c_y-c_h)/s_k, 78 | (c_x+c_w)/s_k, (c_y+c_h)/s_k] 79 | # rest of prior boxes 80 | for ar in self.aspect_ratios[i]: 81 | if not (abs(ar-1) < 1e-6): 82 | c_w = self.min_sizes[i] * sqrt(ar)/2 83 | c_h = self.min_sizes[i] / sqrt(ar)/2 84 | mean += [(c_x-c_w)/s_k, (c_y-c_h)/s_k, 85 | (c_x+c_w)/s_k, (c_y+c_h)/s_k] 86 | # back to torch land 87 | output = torch.Tensor(mean).view(-1, 4) 88 | if self.clip: 89 | output.clamp_(max=1, min=0) 90 | return output 91 | -------------------------------------------------------------------------------- /layers/modules/__init__.py: -------------------------------------------------------------------------------- 1 | from .l2norm import L2Norm 2 | from .multibox_loss import MultiBoxLoss 3 | 4 | __all__ = ['L2Norm', 'MultiBoxLoss'] 5 | -------------------------------------------------------------------------------- /layers/modules/l2norm.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import torch.nn as nn 3 | from torch.autograd import Function 4 | from torch.autograd import Variable 5 | import torch.nn.init as init 6 | 7 | class L2Norm(nn.Module): 8 | def __init__(self,n_channels, scale): 9 | super(L2Norm,self).__init__() 10 | self.n_channels = n_channels 11 | self.gamma = scale or None 12 | self.eps = 1e-10 13 | self.weight = nn.Parameter(torch.Tensor(self.n_channels)) 14 | self.reset_parameters() 15 | 16 | def reset_parameters(self): 17 | init.constant(self.weight,self.gamma) 18 | 19 | def forward(self, x): 20 | norm = x.pow(2).sum(dim=1, keepdim=True).sqrt()+self.eps 21 | #x /= norm 22 | x = torch.div(x,norm) 23 | out = self.weight.unsqueeze(0).unsqueeze(2).unsqueeze(3).expand_as(x) * x 24 | return out 25 | -------------------------------------------------------------------------------- /lib/datasets/VOCdevkit-matlab-wrapper/get_voc_opts.m: -------------------------------------------------------------------------------- 1 | function VOCopts = get_voc_opts(path) 2 | 3 | tmp = pwd; 4 | cd(path); 5 | try 6 | addpath('VOCcode'); 7 | VOCinit; 8 | catch 9 | rmpath('VOCcode'); 10 | cd(tmp); 11 | error(sprintf('VOCcode directory not found under %s', path)); 12 | end 13 | rmpath('VOCcode'); 14 | cd(tmp); 15 | -------------------------------------------------------------------------------- /lib/datasets/VOCdevkit-matlab-wrapper/voc_eval.m: -------------------------------------------------------------------------------- 1 | function res = voc_eval(path, comp_id, test_set, output_dir) 2 | 3 | VOCopts = get_voc_opts(path); 4 | VOCopts.testset = test_set; 5 | 6 | for i = 1:length(VOCopts.classes) 7 | cls = VOCopts.classes{i}; 8 | res(i) = voc_eval_cls(cls, VOCopts, comp_id, output_dir); 9 | end 10 | 11 | fprintf('\n~~~~~~~~~~~~~~~~~~~~\n'); 12 | fprintf('Results:\n'); 13 | aps = [res(:).ap]'; 14 | fprintf('%.1f\n', aps * 100); 15 | fprintf('%.1f\n', mean(aps) * 100); 16 | fprintf('~~~~~~~~~~~~~~~~~~~~\n'); 17 | 18 | function res = voc_eval_cls(cls, VOCopts, comp_id, output_dir) 19 | 20 | test_set = VOCopts.testset; 21 | year = VOCopts.dataset(4:end); 22 | 23 | addpath(fullfile(VOCopts.datadir, 'VOCcode')); 24 | 25 | res_fn = sprintf(VOCopts.detrespath, comp_id, cls); 26 | 27 | recall = []; 28 | prec = []; 29 | ap = 0; 30 | ap_auc = 0; 31 | 32 | do_eval = (str2num(year) <= 2007) | ~strcmp(test_set, 'test'); 33 | if do_eval 34 | % Bug in VOCevaldet requires that tic has been called first 35 | tic; 36 | [recall, prec, ap] = VOCevaldet(VOCopts, comp_id, cls, true); 37 | ap_auc = xVOCap(recall, prec); 38 | 39 | % force plot limits 40 | ylim([0 1]); 41 | xlim([0 1]); 42 | 43 | print(gcf, '-djpeg', '-r0', ... 44 | [output_dir '/' cls '_pr.jpg']); 45 | end 46 | fprintf('!!! %s : %.4f %.4f\n', cls, ap, ap_auc); 47 | 48 | res.recall = recall; 49 | res.prec = prec; 50 | res.ap = ap; 51 | res.ap_auc = ap_auc; 52 | 53 | save([output_dir '/' cls '_pr.mat'], ... 54 | 'res', 'recall', 'prec', 'ap', 'ap_auc'); 55 | 56 | rmpath(fullfile(VOCopts.datadir, 'VOCcode')); 57 | -------------------------------------------------------------------------------- /lib/datasets/VOCdevkit-matlab-wrapper/xVOCap.m: -------------------------------------------------------------------------------- 1 | function ap = xVOCap(rec,prec) 2 | % From the PASCAL VOC 2011 devkit 3 | 4 | mrec=[0 ; rec ; 1]; 5 | mpre=[0 ; prec ; 0]; 6 | for i=numel(mpre)-1:-1:1 7 | mpre(i)=max(mpre(i),mpre(i+1)); 8 | end 9 | i=find(mrec(2:end)~=mrec(1:end-1))+1; 10 | ap=sum((mrec(i)-mrec(i-1)).*mpre(i)); 11 | -------------------------------------------------------------------------------- /lib/datasets/__init__.py: -------------------------------------------------------------------------------- 1 | # -------------------------------------------------------- 2 | # Fast R-CNN 3 | # Copyright (c) 2015 Microsoft 4 | # Licensed under The MIT License [see LICENSE for details] 5 | # Written by Ross Girshick 6 | # -------------------------------------------------------- 7 | -------------------------------------------------------------------------------- /lib/datasets/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/__init__.pyc -------------------------------------------------------------------------------- /lib/datasets/__pycache__/__init__.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/__pycache__/__init__.cpython-36.pyc -------------------------------------------------------------------------------- /lib/datasets/coco.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/coco.pyc -------------------------------------------------------------------------------- /lib/datasets/ds_utils.py: -------------------------------------------------------------------------------- 1 | # -------------------------------------------------------- 2 | # Fast/er R-CNN 3 | # Licensed under The MIT License [see LICENSE for details] 4 | # Written by Ross Girshick 5 | # -------------------------------------------------------- 6 | from __future__ import absolute_import 7 | from __future__ import division 8 | from __future__ import print_function 9 | 10 | import numpy as np 11 | 12 | 13 | def unique_boxes(boxes, scale=1.0): 14 | """Return indices of unique boxes.""" 15 | v = np.array([1, 1e3, 1e6, 1e9]) 16 | hashes = np.round(boxes * scale).dot(v) 17 | _, index = np.unique(hashes, return_index=True) 18 | return np.sort(index) 19 | 20 | 21 | def xywh_to_xyxy(boxes): 22 | """Convert [x y w h] box format to [x1 y1 x2 y2] format.""" 23 | return np.hstack((boxes[:, 0:2], boxes[:, 0:2] + boxes[:, 2:4] - 1)) 24 | 25 | 26 | def xyxy_to_xywh(boxes): 27 | """Convert [x1 y1 x2 y2] box format to [x y w h] format.""" 28 | return np.hstack((boxes[:, 0:2], boxes[:, 2:4] - boxes[:, 0:2] + 1)) 29 | 30 | 31 | def validate_boxes(boxes, width=0, height=0): 32 | """Check that a set of boxes are valid.""" 33 | x1 = boxes[:, 0] 34 | y1 = boxes[:, 1] 35 | x2 = boxes[:, 2] 36 | y2 = boxes[:, 3] 37 | assert (x1 >= 0).all() 38 | assert (y1 >= 0).all() 39 | assert (x2 >= x1).all() 40 | assert (y2 >= y1).all() 41 | assert (x2 < width).all() 42 | assert (y2 < height).all() 43 | 44 | 45 | def filter_small_boxes(boxes, min_size): 46 | w = boxes[:, 2] - boxes[:, 0] 47 | h = boxes[:, 3] - boxes[:, 1] 48 | keep = np.where((w >= min_size) & (h > min_size))[0] 49 | return keep 50 | -------------------------------------------------------------------------------- /lib/datasets/ds_utils.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/ds_utils.pyc -------------------------------------------------------------------------------- /lib/datasets/factory.py: -------------------------------------------------------------------------------- 1 | # -------------------------------------------------------- 2 | # Fast R-CNN 3 | # Copyright (c) 2015 Microsoft 4 | # Licensed under The MIT License [see LICENSE for details] 5 | # Written by Ross Girshick 6 | # -------------------------------------------------------- 7 | 8 | """Factory method for easily getting imdbs by name.""" 9 | from __future__ import absolute_import 10 | from __future__ import division 11 | from __future__ import print_function 12 | 13 | __sets = {} 14 | from datasets.pascal_voc import pascal_voc 15 | from datasets.coco import coco 16 | from datasets.imagenet import imagenet 17 | from datasets.vg import vg 18 | 19 | import numpy as np 20 | 21 | # Set up voc__ 22 | for year in ['2007', '2012']: 23 | for split in ['train', 'val', 'trainval', 'test']: 24 | name = 'voc_{}_{}'.format(year, split) 25 | __sets[name] = (lambda split=split, year=year: pascal_voc(split, year)) 26 | 27 | # Set up coco_2014_ 28 | for year in ['2014']: 29 | for split in ['train', 'val', 'minival', 'valminusminival', 'trainval']: 30 | name = 'coco_{}_{}'.format(year, split) 31 | __sets[name] = (lambda split=split, year=year: coco(split, year)) 32 | 33 | # Set up coco_2014_cap_ 34 | for year in ['2014']: 35 | for split in ['train', 'val', 'capval', 'valminuscapval', 'trainval']: 36 | name = 'coco_{}_{}'.format(year, split) 37 | __sets[name] = (lambda split=split, year=year: coco(split, year)) 38 | 39 | # Set up coco_2015_ 40 | for year in ['2015']: 41 | for split in ['test', 'test-dev']: 42 | name = 'coco_{}_{}'.format(year, split) 43 | __sets[name] = (lambda split=split, year=year: coco(split, year)) 44 | 45 | # Set up vg_ 46 | # for version in ['1600-400-20']: 47 | # for split in ['minitrain', 'train', 'minival', 'val', 'test']: 48 | # name = 'vg_{}_{}'.format(version,split) 49 | # __sets[name] = (lambda split=split, version=version: vg(version, split)) 50 | for version in ['150-50-20', '150-50-50', '500-150-80', '750-250-150', '1750-700-450', '1600-400-20']: 51 | for split in ['minitrain', 'smalltrain', 'train', 'minival', 'smallval', 'val', 'test']: 52 | name = 'vg_{}_{}'.format(version,split) 53 | __sets[name] = (lambda split=split, version=version: vg(version, split)) 54 | 55 | # set up image net. 56 | for split in ['train', 'val', 'val1', 'val2', 'test']: 57 | name = 'imagenet_{}'.format(split) 58 | devkit_path = 'data/imagenet/ILSVRC/devkit' 59 | data_path = 'data/imagenet/ILSVRC' 60 | __sets[name] = (lambda split=split, devkit_path=devkit_path, data_path=data_path: imagenet(split,devkit_path,data_path)) 61 | 62 | def get_imdb(name): 63 | """Get an imdb (image database) by name.""" 64 | if name not in __sets: 65 | raise KeyError('Unknown dataset: {}'.format(name)) 66 | return __sets[name]() 67 | 68 | 69 | def list_imdbs(): 70 | """List all registered imdbs.""" 71 | return list(__sets.keys()) 72 | -------------------------------------------------------------------------------- /lib/datasets/factory.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/factory.pyc -------------------------------------------------------------------------------- /lib/datasets/imagenet.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/imagenet.pyc -------------------------------------------------------------------------------- /lib/datasets/imdb.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/imdb.pyc -------------------------------------------------------------------------------- /lib/datasets/pascal_voc.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/pascal_voc.pyc -------------------------------------------------------------------------------- /lib/datasets/tools/mcg_munge.py: -------------------------------------------------------------------------------- 1 | from __future__ import print_function 2 | import os 3 | import sys 4 | 5 | """Hacky tool to convert file system layout of MCG boxes downloaded from 6 | http://www.eecs.berkeley.edu/Research/Projects/CS/vision/grouping/mcg/ 7 | so that it's consistent with those computed by Jan Hosang (see: 8 | http://www.mpi-inf.mpg.de/departments/computer-vision-and-multimodal- 9 | computing/research/object-recognition-and-scene-understanding/how- 10 | good-are-detection-proposals-really/) 11 | 12 | NB: Boxes from the MCG website are in (y1, x1, y2, x2) order. 13 | Boxes from Hosang et al. are in (x1, y1, x2, y2) order. 14 | """ 15 | 16 | def munge(src_dir): 17 | # stored as: ./MCG-COCO-val2014-boxes/COCO_val2014_000000193401.mat 18 | # want: ./MCG/mat/COCO_val2014_0/COCO_val2014_000000141/COCO_val2014_000000141334.mat 19 | 20 | files = os.listdir(src_dir) 21 | for fn in files: 22 | base, ext = os.path.splitext(fn) 23 | # first 14 chars / first 22 chars / all chars + .mat 24 | # COCO_val2014_0/COCO_val2014_000000447/COCO_val2014_000000447991.mat 25 | first = base[:14] 26 | second = base[:22] 27 | dst_dir = os.path.join('MCG', 'mat', first, second) 28 | if not os.path.exists(dst_dir): 29 | os.makedirs(dst_dir) 30 | src = os.path.join(src_dir, fn) 31 | dst = os.path.join(dst_dir, fn) 32 | print('MV: {} -> {}'.format(src, dst)) 33 | os.rename(src, dst) 34 | 35 | if __name__ == '__main__': 36 | # src_dir should look something like: 37 | # src_dir = 'MCG-COCO-val2014-boxes' 38 | src_dir = sys.argv[1] 39 | munge(src_dir) 40 | -------------------------------------------------------------------------------- /lib/datasets/vg.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/vg.pyc -------------------------------------------------------------------------------- /lib/datasets/vg_eval.py: -------------------------------------------------------------------------------- 1 | from __future__ import absolute_import 2 | # -------------------------------------------------------- 3 | # Fast/er R-CNN 4 | # Licensed under The MIT License [see LICENSE for details] 5 | # Written by Bharath Hariharan 6 | # -------------------------------------------------------- 7 | 8 | import xml.etree.ElementTree as ET 9 | import os 10 | import numpy as np 11 | from .voc_eval import voc_ap 12 | 13 | def vg_eval( detpath, 14 | gt_roidb, 15 | image_index, 16 | classindex, 17 | ovthresh=0.5, 18 | use_07_metric=False, 19 | eval_attributes=False): 20 | """rec, prec, ap, sorted_scores, npos = voc_eval( 21 | detpath, 22 | gt_roidb, 23 | image_index, 24 | classindex, 25 | [ovthresh], 26 | [use_07_metric]) 27 | 28 | Top level function that does the Visual Genome evaluation. 29 | 30 | detpath: Path to detections 31 | gt_roidb: List of ground truth structs. 32 | image_index: List of image ids. 33 | classindex: Category index 34 | [ovthresh]: Overlap threshold (default = 0.5) 35 | [use_07_metric]: Whether to use VOC07's 11 point AP computation 36 | (default False) 37 | """ 38 | # extract gt objects for this class 39 | class_recs = {} 40 | npos = 0 41 | for item,imagename in zip(gt_roidb,image_index): 42 | if eval_attributes: 43 | bbox = item['boxes'][np.where(np.any(item['gt_attributes'].toarray() == classindex, axis=1))[0], :] 44 | else: 45 | bbox = item['boxes'][np.where(item['gt_classes'] == classindex)[0], :] 46 | difficult = np.zeros((bbox.shape[0],)).astype(np.bool) 47 | det = [False] * bbox.shape[0] 48 | npos = npos + sum(~difficult) 49 | class_recs[str(imagename)] = {'bbox': bbox, 50 | 'difficult': difficult, 51 | 'det': det} 52 | if npos == 0: 53 | # No ground truth examples 54 | return 0,0,0,0,npos 55 | 56 | # read dets 57 | with open(detpath, 'r') as f: 58 | lines = f.readlines() 59 | if len(lines) == 0: 60 | # No detection examples 61 | return 0,0,0,0,npos 62 | 63 | splitlines = [x.strip().split(' ') for x in lines] 64 | image_ids = [x[0] for x in splitlines] 65 | confidence = np.array([float(x[1]) for x in splitlines]) 66 | BB = np.array([[float(z) for z in x[2:]] for x in splitlines]) 67 | 68 | # sort by confidence 69 | sorted_ind = np.argsort(-confidence) 70 | sorted_scores = -np.sort(-confidence) 71 | BB = BB[sorted_ind, :] 72 | image_ids = [image_ids[x] for x in sorted_ind] 73 | 74 | # go down dets and mark TPs and FPs 75 | nd = len(image_ids) 76 | tp = np.zeros(nd) 77 | fp = np.zeros(nd) 78 | for d in range(nd): 79 | R = class_recs[image_ids[d]] 80 | bb = BB[d, :].astype(float) 81 | ovmax = -np.inf 82 | BBGT = R['bbox'].astype(float) 83 | 84 | if BBGT.size > 0: 85 | # compute overlaps 86 | # intersection 87 | ixmin = np.maximum(BBGT[:, 0], bb[0]) 88 | iymin = np.maximum(BBGT[:, 1], bb[1]) 89 | ixmax = np.minimum(BBGT[:, 2], bb[2]) 90 | iymax = np.minimum(BBGT[:, 3], bb[3]) 91 | iw = np.maximum(ixmax - ixmin + 1., 0.) 92 | ih = np.maximum(iymax - iymin + 1., 0.) 93 | inters = iw * ih 94 | 95 | # union 96 | uni = ((bb[2] - bb[0] + 1.) * (bb[3] - bb[1] + 1.) + 97 | (BBGT[:, 2] - BBGT[:, 0] + 1.) * 98 | (BBGT[:, 3] - BBGT[:, 1] + 1.) - inters) 99 | 100 | overlaps = inters / uni 101 | ovmax = np.max(overlaps) 102 | jmax = np.argmax(overlaps) 103 | 104 | if ovmax > ovthresh: 105 | if not R['difficult'][jmax]: 106 | if not R['det'][jmax]: 107 | tp[d] = 1. 108 | R['det'][jmax] = 1 109 | else: 110 | fp[d] = 1. 111 | else: 112 | fp[d] = 1. 113 | 114 | # compute precision recall 115 | fp = np.cumsum(fp) 116 | tp = np.cumsum(tp) 117 | rec = tp / float(npos) 118 | # avoid divide by zero in case the first detection matches a difficult 119 | # ground truth 120 | prec = tp / np.maximum(tp + fp, np.finfo(np.float64).eps) 121 | ap = voc_ap(rec, prec, use_07_metric) 122 | 123 | return rec, prec, ap, sorted_scores, npos 124 | -------------------------------------------------------------------------------- /lib/datasets/vg_eval.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/vg_eval.pyc -------------------------------------------------------------------------------- /lib/datasets/voc_eval.py: -------------------------------------------------------------------------------- 1 | # -------------------------------------------------------- 2 | # Fast/er R-CNN 3 | # Licensed under The MIT License [see LICENSE for details] 4 | # Written by Bharath Hariharan 5 | # -------------------------------------------------------- 6 | from __future__ import absolute_import 7 | from __future__ import division 8 | from __future__ import print_function 9 | 10 | import xml.etree.ElementTree as ET 11 | import os 12 | import pickle 13 | import numpy as np 14 | 15 | def parse_rec(filename): 16 | """ Parse a PASCAL VOC xml file """ 17 | tree = ET.parse(filename) 18 | objects = [] 19 | for obj in tree.findall('object'): 20 | obj_struct = {} 21 | obj_struct['name'] = obj.find('name').text 22 | obj_struct['pose'] = obj.find('pose').text 23 | obj_struct['truncated'] = int(obj.find('truncated').text) 24 | obj_struct['difficult'] = int(obj.find('difficult').text) 25 | bbox = obj.find('bndbox') 26 | obj_struct['bbox'] = [int(bbox.find('xmin').text), 27 | int(bbox.find('ymin').text), 28 | int(bbox.find('xmax').text), 29 | int(bbox.find('ymax').text)] 30 | objects.append(obj_struct) 31 | 32 | return objects 33 | 34 | 35 | def voc_ap(rec, prec, use_07_metric=False): 36 | """ ap = voc_ap(rec, prec, [use_07_metric]) 37 | Compute VOC AP given precision and recall. 38 | If use_07_metric is true, uses the 39 | VOC 07 11 point method (default:False). 40 | """ 41 | if use_07_metric: 42 | # 11 point metric 43 | ap = 0. 44 | for t in np.arange(0., 1.1, 0.1): 45 | if np.sum(rec >= t) == 0: 46 | p = 0 47 | else: 48 | p = np.max(prec[rec >= t]) 49 | ap = ap + p / 11. 50 | else: 51 | # correct AP calculation 52 | # first append sentinel values at the end 53 | mrec = np.concatenate(([0.], rec, [1.])) 54 | mpre = np.concatenate(([0.], prec, [0.])) 55 | 56 | # compute the precision envelope 57 | for i in range(mpre.size - 1, 0, -1): 58 | mpre[i - 1] = np.maximum(mpre[i - 1], mpre[i]) 59 | 60 | # to calculate area under PR curve, look for points 61 | # where X axis (recall) changes value 62 | i = np.where(mrec[1:] != mrec[:-1])[0] 63 | 64 | # and sum (\Delta recall) * prec 65 | ap = np.sum((mrec[i + 1] - mrec[i]) * mpre[i + 1]) 66 | return ap 67 | 68 | 69 | def voc_eval(detpath, 70 | annopath, 71 | imagesetfile, 72 | classname, 73 | cachedir, 74 | ovthresh=0.5, 75 | use_07_metric=False): 76 | """rec, prec, ap = voc_eval(detpath, 77 | annopath, 78 | imagesetfile, 79 | classname, 80 | [ovthresh], 81 | [use_07_metric]) 82 | 83 | Top level function that does the PASCAL VOC evaluation. 84 | 85 | detpath: Path to detections 86 | detpath.format(classname) should produce the detection results file. 87 | annopath: Path to annotations 88 | annopath.format(imagename) should be the xml annotations file. 89 | imagesetfile: Text file containing the list of images, one image per line. 90 | classname: Category name (duh) 91 | cachedir: Directory for caching the annotations 92 | [ovthresh]: Overlap threshold (default = 0.5) 93 | [use_07_metric]: Whether to use VOC07's 11 point AP computation 94 | (default False) 95 | """ 96 | # assumes detections are in detpath.format(classname) 97 | # assumes annotations are in annopath.format(imagename) 98 | # assumes imagesetfile is a text file with each line an image name 99 | # cachedir caches the annotations in a pickle file 100 | 101 | # first load gt 102 | if not os.path.isdir(cachedir): 103 | os.mkdir(cachedir) 104 | cachefile = os.path.join(cachedir, '%s_annots.pkl' % imagesetfile) 105 | # read list of images 106 | with open(imagesetfile, 'r') as f: 107 | lines = f.readlines() 108 | imagenames = [x.strip() for x in lines] 109 | 110 | if not os.path.isfile(cachefile): 111 | # load annotations 112 | recs = {} 113 | for i, imagename in enumerate(imagenames): 114 | recs[imagename] = parse_rec(annopath.format(imagename)) 115 | if i % 100 == 0: 116 | print('Reading annotation for {:d}/{:d}'.format( 117 | i + 1, len(imagenames))) 118 | # save 119 | print('Saving cached annotations to {:s}'.format(cachefile)) 120 | with open(cachefile, 'wb') as f: 121 | pickle.dump(recs, f) 122 | else: 123 | # load 124 | with open(cachefile, 'rb') as f: 125 | try: 126 | recs = pickle.load(f) 127 | except: 128 | recs = pickle.load(f, encoding='bytes') 129 | 130 | # extract gt objects for this class 131 | class_recs = {} 132 | npos = 0 133 | for imagename in imagenames: 134 | R = [obj for obj in recs[imagename] if obj['name'] == classname] 135 | bbox = np.array([x['bbox'] for x in R]) 136 | difficult = np.array([x['difficult'] for x in R]).astype(np.bool) 137 | det = [False] * len(R) 138 | npos = npos + sum(~difficult) 139 | class_recs[imagename] = {'bbox': bbox, 140 | 'difficult': difficult, 141 | 'det': det} 142 | 143 | # read dets 144 | detfile = detpath.format(classname) 145 | with open(detfile, 'r') as f: 146 | lines = f.readlines() 147 | 148 | splitlines = [x.strip().split(' ') for x in lines] 149 | image_ids = [x[0] for x in splitlines] 150 | confidence = np.array([float(x[1]) for x in splitlines]) 151 | BB = np.array([[float(z) for z in x[2:]] for x in splitlines]) 152 | 153 | nd = len(image_ids) 154 | tp = np.zeros(nd) 155 | fp = np.zeros(nd) 156 | 157 | if BB.shape[0] > 0: 158 | # sort by confidence 159 | sorted_ind = np.argsort(-confidence) 160 | sorted_scores = np.sort(-confidence) 161 | BB = BB[sorted_ind, :] 162 | image_ids = [image_ids[x] for x in sorted_ind] 163 | 164 | # go down dets and mark TPs and FPs 165 | for d in range(nd): 166 | R = class_recs[image_ids[d]] 167 | bb = BB[d, :].astype(float) 168 | ovmax = -np.inf 169 | BBGT = R['bbox'].astype(float) 170 | 171 | if BBGT.size > 0: 172 | # compute overlaps 173 | # intersection 174 | ixmin = np.maximum(BBGT[:, 0], bb[0]) 175 | iymin = np.maximum(BBGT[:, 1], bb[1]) 176 | ixmax = np.minimum(BBGT[:, 2], bb[2]) 177 | iymax = np.minimum(BBGT[:, 3], bb[3]) 178 | iw = np.maximum(ixmax - ixmin + 1., 0.) 179 | ih = np.maximum(iymax - iymin + 1., 0.) 180 | inters = iw * ih 181 | 182 | # union 183 | uni = ((bb[2] - bb[0] + 1.) * (bb[3] - bb[1] + 1.) + 184 | (BBGT[:, 2] - BBGT[:, 0] + 1.) * 185 | (BBGT[:, 3] - BBGT[:, 1] + 1.) - inters) 186 | 187 | overlaps = inters / uni 188 | ovmax = np.max(overlaps) 189 | jmax = np.argmax(overlaps) 190 | 191 | if ovmax > ovthresh: 192 | if not R['difficult'][jmax]: 193 | if not R['det'][jmax]: 194 | tp[d] = 1. 195 | R['det'][jmax] = 1 196 | else: 197 | fp[d] = 1. 198 | else: 199 | fp[d] = 1. 200 | 201 | # compute precision recall 202 | fp = np.cumsum(fp) 203 | tp = np.cumsum(tp) 204 | rec = tp / float(npos) 205 | # avoid divide by zero in case the first detection matches a difficult 206 | # ground truth 207 | prec = tp / np.maximum(tp + fp, np.finfo(np.float64).eps) 208 | ap = voc_ap(rec, prec, use_07_metric) 209 | 210 | return rec, prec, ap 211 | -------------------------------------------------------------------------------- /lib/datasets/voc_eval.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/datasets/voc_eval.pyc -------------------------------------------------------------------------------- /lib/make.sh: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env bash 2 | 3 | CUDA_PATH=/usr/local/cuda/ 4 | 5 | python setup.py build_ext --inplace 6 | rm -rf build 7 | 8 | CUDA_ARCH="-gencode arch=compute_30,code=sm_61" 9 | 10 | # compile NMS 11 | cd model/nms/src 12 | echo "Compiling nms kernels by nvcc..." 13 | nvcc -c -o nms_cuda_kernel.cu.o nms_cuda_kernel.cu \ 14 | -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH 15 | 16 | cd ../ 17 | python build.py 18 | 19 | # compile roi_pooling 20 | cd ../../ 21 | cd model/roi_pooling/src 22 | echo "Compiling roi pooling kernels by nvcc..." 23 | nvcc -c -o roi_pooling.cu.o roi_pooling_kernel.cu \ 24 | -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH 25 | cd ../ 26 | python build.py 27 | 28 | # compile roi_align 29 | cd ../../ 30 | cd model/roi_align/src 31 | echo "Compiling roi align kernels by nvcc..." 32 | nvcc -c -o roi_align_kernel.cu.o roi_align_kernel.cu \ 33 | -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH 34 | cd ../ 35 | python build.py 36 | 37 | # compile roi_crop 38 | cd ../../ 39 | cd model/roi_crop/src 40 | echo "Compiling roi crop kernels by nvcc..." 41 | nvcc -c -o roi_crop_cuda_kernel.cu.o roi_crop_cuda_kernel.cu \ 42 | -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH 43 | cd ../ 44 | python build.py 45 | -------------------------------------------------------------------------------- /lib/model/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/__init__.py -------------------------------------------------------------------------------- /lib/model/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/__init__.pyc -------------------------------------------------------------------------------- /lib/model/__pycache__/__init__.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/__pycache__/__init__.cpython-36.pyc -------------------------------------------------------------------------------- /lib/model/faster_rcnn/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/faster_rcnn/__init__.py -------------------------------------------------------------------------------- /lib/model/faster_rcnn/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/faster_rcnn/__init__.pyc -------------------------------------------------------------------------------- /lib/model/faster_rcnn/faster_rcnn.py: -------------------------------------------------------------------------------- 1 | import random 2 | import torch 3 | import torch.nn as nn 4 | import torch.nn.functional as F 5 | from torch.autograd import Variable 6 | import torchvision.models as models 7 | from torch.autograd import Variable 8 | import numpy as np 9 | from model.utils.config import cfg 10 | from model.rpn.rpn import _RPN 11 | from model.roi_pooling.modules.roi_pool import _RoIPooling 12 | from model.roi_crop.modules.roi_crop import _RoICrop 13 | from model.roi_align.modules.roi_align import RoIAlignAvg 14 | from model.rpn.proposal_target_layer_cascade import _ProposalTargetLayer 15 | import time 16 | import pdb 17 | from model.utils.net_utils import _smooth_l1_loss, _crop_pool_layer, _affine_grid_gen, _affine_theta 18 | from model.roi_pooling_replace import roi_pooling 19 | 20 | class _fasterRCNN(nn.Module): 21 | """ faster RCNN """ 22 | def __init__(self, classes, class_agnostic): 23 | super(_fasterRCNN, self).__init__() 24 | self.classes = classes 25 | self.n_classes = len(classes) 26 | self.class_agnostic = class_agnostic 27 | # loss 28 | self.RCNN_loss_cls = 0 29 | self.RCNN_loss_bbox = 0 30 | 31 | # define rpn 32 | self.RCNN_rpn = _RPN(self.dout_base_model) 33 | self.RCNN_proposal_target = _ProposalTargetLayer(self.n_classes) 34 | self.RCNN_roi_pool = _RoIPooling(cfg.POOLING_SIZE, cfg.POOLING_SIZE, 1.0/16.0) 35 | self.RCNN_roi_align = RoIAlignAvg(cfg.POOLING_SIZE, cfg.POOLING_SIZE, 1.0/16.0) 36 | 37 | self.grid_size = cfg.POOLING_SIZE * 2 if cfg.CROP_RESIZE_WITH_MAX_POOL else cfg.POOLING_SIZE 38 | self.RCNN_roi_crop = _RoICrop() 39 | 40 | def forward(self, im_data, im_info, gt_boxes, num_boxes): 41 | batch_size = im_data.size(0) 42 | 43 | im_info = im_info.data 44 | gt_boxes = gt_boxes.data 45 | num_boxes = num_boxes.data 46 | 47 | # feed image data to base model to obtain base feature map 48 | base_feat = self.RCNN_base(im_data) 49 | 50 | # feed base feature map tp RPN to obtain rois 51 | rois, rpn_loss_cls, rpn_loss_bbox = self.RCNN_rpn(base_feat, im_info, gt_boxes, num_boxes) 52 | 53 | # if it is training phrase, then use ground trubut bboxes for refining 54 | if self.training: 55 | roi_data = self.RCNN_proposal_target(rois, gt_boxes, num_boxes) 56 | rois, rois_label, rois_target, rois_inside_ws, rois_outside_ws = roi_data 57 | 58 | rois_label = Variable(rois_label.view(-1).long()) 59 | rois_target = Variable(rois_target.view(-1, rois_target.size(2))) 60 | rois_inside_ws = Variable(rois_inside_ws.view(-1, rois_inside_ws.size(2))) 61 | rois_outside_ws = Variable(rois_outside_ws.view(-1, rois_outside_ws.size(2))) 62 | else: 63 | rois_label = None 64 | rois_target = None 65 | rois_inside_ws = None 66 | rois_outside_ws = None 67 | rpn_loss_cls = 0 68 | rpn_loss_bbox = 0 69 | 70 | rois = Variable(rois) 71 | # do roi pooling based on predicted rois 72 | 73 | if cfg.POOLING_MODE == 'crop': 74 | # pdb.set_trace() 75 | # pooled_feat_anchor = _crop_pool_layer(base_feat, rois.view(-1, 5)) 76 | grid_xy = _affine_grid_gen(rois.view(-1, 5), base_feat.size()[2:], self.grid_size) 77 | grid_yx = torch.stack([grid_xy.data[:,:,:,1], grid_xy.data[:,:,:,0]], 3).contiguous() 78 | pooled_feat = self.RCNN_roi_crop(base_feat, Variable(grid_yx).detach()) 79 | if cfg.CROP_RESIZE_WITH_MAX_POOL: 80 | pooled_feat = F.max_pool2d(pooled_feat, 2, 2) 81 | elif cfg.POOLING_MODE == 'align': 82 | pooled_feat = self.RCNN_roi_align(base_feat, rois.view(-1, 5)) 83 | elif cfg.POOLING_MODE == 'pool': 84 | print(rois) 85 | pooled_feat = self.RCNN_roi_pool(base_feat, rois.view(-1,5)) 86 | #print(rois.size()) 87 | #print(torch.clamp(rois.view(-1,5)[:,1:], 0, 1).size()) 88 | ''' 89 | rois = rois.view(-1,5) 90 | rois[:,1:] = torch.clamp(rois[:,1:], 0, 1) 91 | pooled_feat = roi_pooling(base_feat, rois, (cfg.POOLING_SIZE, cfg.POOLING_SIZE), 1.0/16.0) 92 | rois = rois.view(1,-1,5) 93 | ''' 94 | # feed pooled features to top model 95 | pooled_feat = self._head_to_tail(pooled_feat) 96 | 97 | # compute bbox offset 98 | bbox_pred = self.RCNN_bbox_pred(pooled_feat) 99 | if self.training and not self.class_agnostic: 100 | # select the corresponding columns according to roi labels 101 | bbox_pred_view = bbox_pred.view(bbox_pred.size(0), int(bbox_pred.size(1) / 4), 4) 102 | bbox_pred_select = torch.gather(bbox_pred_view, 1, rois_label.view(rois_label.size(0), 1, 1).expand(rois_label.size(0), 1, 4)) 103 | bbox_pred = bbox_pred_select.squeeze(1) 104 | 105 | # compute object classification probability 106 | cls_score = self.RCNN_cls_score(pooled_feat) 107 | cls_prob = F.softmax(cls_score) 108 | 109 | RCNN_loss_cls = 0 110 | RCNN_loss_bbox = 0 111 | 112 | if self.training: 113 | # classification loss 114 | RCNN_loss_cls = F.cross_entropy(cls_score, rois_label) 115 | 116 | # bounding box regression L1 loss 117 | RCNN_loss_bbox = _smooth_l1_loss(bbox_pred, rois_target, rois_inside_ws, rois_outside_ws) 118 | 119 | 120 | cls_prob = cls_prob.view(batch_size, rois.size(1), -1) 121 | bbox_pred = bbox_pred.view(batch_size, rois.size(1), -1) 122 | 123 | return rois, cls_prob, bbox_pred, rpn_loss_cls, rpn_loss_bbox, RCNN_loss_cls, RCNN_loss_bbox, rois_label 124 | 125 | def _init_weights(self): 126 | def normal_init(m, mean, stddev, truncated=False): 127 | """ 128 | weight initalizer: truncated normal and random normal. 129 | """ 130 | # x is a parameter 131 | if truncated: 132 | m.weight.data.normal_().fmod_(2).mul_(stddev).add_(mean) # not a perfect approximation 133 | else: 134 | m.weight.data.normal_(mean, stddev) 135 | m.bias.data.zero_() 136 | 137 | normal_init(self.RCNN_rpn.RPN_Conv, 0, 0.01, cfg.TRAIN.TRUNCATED) 138 | normal_init(self.RCNN_rpn.RPN_cls_score, 0, 0.01, cfg.TRAIN.TRUNCATED) 139 | normal_init(self.RCNN_rpn.RPN_bbox_pred, 0, 0.01, cfg.TRAIN.TRUNCATED) 140 | normal_init(self.RCNN_cls_score, 0, 0.01, cfg.TRAIN.TRUNCATED) 141 | normal_init(self.RCNN_bbox_pred, 0, 0.001, cfg.TRAIN.TRUNCATED) 142 | 143 | def create_architecture(self): 144 | self._init_modules() 145 | self._init_weights() 146 | -------------------------------------------------------------------------------- /lib/model/faster_rcnn/faster_rcnn.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/faster_rcnn/faster_rcnn.pyc -------------------------------------------------------------------------------- /lib/model/faster_rcnn/resnet.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/faster_rcnn/resnet.pyc -------------------------------------------------------------------------------- /lib/model/faster_rcnn/vgg16.py: -------------------------------------------------------------------------------- 1 | # -------------------------------------------------------- 2 | # Tensorflow Faster R-CNN 3 | # Licensed under The MIT License [see LICENSE for details] 4 | # Written by Xinlei Chen 5 | # -------------------------------------------------------- 6 | from __future__ import absolute_import 7 | from __future__ import division 8 | from __future__ import print_function 9 | 10 | import torch 11 | import torch.nn as nn 12 | import torch.nn.functional as F 13 | from torch.autograd import Variable 14 | import math 15 | import torchvision.models as models 16 | from model.faster_rcnn.faster_rcnn import _fasterRCNN 17 | from model.fixed_proj import HadamardProj, Proj 18 | import pdb 19 | 20 | class vgg16(_fasterRCNN): 21 | def __init__(self, classes, pretrained=False, class_agnostic=False): 22 | self.model_path = 'data/pretrained_model/vgg16_caffe.pth' 23 | self.dout_base_model = 512 24 | self.pretrained = pretrained 25 | self.class_agnostic = class_agnostic 26 | 27 | _fasterRCNN.__init__(self, classes, class_agnostic) 28 | 29 | def _init_modules(self): 30 | vgg = models.vgg16() 31 | if self.pretrained: 32 | print("Loading pretrained weights from %s" %(self.model_path)) 33 | state_dict = torch.load(self.model_path) 34 | vgg.load_state_dict({k:v for k,v in state_dict.items() if k in vgg.state_dict()}) 35 | 36 | vgg.classifier = nn.Sequential(*list(vgg.classifier._modules.values())[:-1]) 37 | # not using the last maxpool layer 38 | self.RCNN_base = nn.Sequential(*list(vgg.features._modules.values())[:-1]) 39 | 40 | # Fix the layers before conv3: 41 | for layer in range(10): 42 | for p in self.RCNN_base[layer].parameters(): p.requires_grad = False 43 | 44 | # self.RCNN_base = _RCNN_base(vgg.features, self.classes, self.dout_base_model) 45 | 46 | self.RCNN_top = vgg.classifier 47 | 48 | # not using the last maxpool layer 49 | self.RCNN_cls_score = nn.Linear(4096, self.n_classes) 50 | 51 | #fix top classifier 52 | #self.RCNN_cls_score = Proj(4096, self.n_classes) 53 | 54 | if self.class_agnostic: 55 | self.RCNN_bbox_pred = nn.Linear(4096, 4) 56 | else: 57 | self.RCNN_bbox_pred = nn.Linear(4096, 4 * self.n_classes) 58 | 59 | def _head_to_tail(self, pool5): 60 | pool5_flat = pool5.view(pool5.size(0), -1) 61 | fc7 = self.RCNN_top(pool5_flat) 62 | 63 | return fc7 64 | 65 | -------------------------------------------------------------------------------- /lib/model/faster_rcnn/vgg16.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/faster_rcnn/vgg16.pyc -------------------------------------------------------------------------------- /lib/model/fixed_proj.py: -------------------------------------------------------------------------------- 1 | import torch.nn as nn 2 | import math 3 | import torch 4 | from torch.autograd import Variable 5 | from scipy.linalg import hadamard 6 | 7 | class HadamardProj(nn.Module): 8 | 9 | def __init__(self, input_size, output_size, bias=True, fixed_weights=True, fixed_scale=None): 10 | super(HadamardProj, self).__init__() 11 | self.output_size = output_size 12 | self.input_size = input_size 13 | sz = 2 ** int(math.ceil(math.log(max(input_size, output_size), 2))) 14 | mat = torch.from_numpy(hadamard(sz)) 15 | if fixed_weights: 16 | self.proj = Variable(mat, requires_grad=False) 17 | else: 18 | self.proj = nn.Parameter(mat) 19 | 20 | init_scale = 1. / math.sqrt(self.output_size) 21 | 22 | if fixed_scale is not None: 23 | self.scale = Variable(torch.Tensor( 24 | [fixed_scale]), requires_grad=False) 25 | else: 26 | self.scale = nn.Parameter(torch.Tensor([init_scale])) 27 | 28 | if bias: 29 | self.bias = nn.Parameter(torch.Tensor( 30 | output_size).uniform_(-init_scale, init_scale)) 31 | else: 32 | self.register_parameter('bias', None) 33 | 34 | self.eps = 1e-8 35 | 36 | def forward(self, x): 37 | if not isinstance(self.scale, nn.Parameter): 38 | self.scale = self.scale.type_as(x) 39 | x = x / (x.norm(2, -1, keepdim=True) + self.eps) 40 | w = self.proj.type_as(x) 41 | 42 | out = -self.scale * \ 43 | nn.functional.linear(x, w[:self.output_size, :self.input_size]) 44 | if self.bias is not None: 45 | out = out + self.bias.view(1, -1) 46 | return out 47 | 48 | 49 | class Proj(nn.Module): 50 | 51 | def __init__(self, input_size, output_size, bias=True, init_scale=15): 52 | super(Proj, self).__init__() 53 | if init_scale is not None: 54 | #self.weight = nn.Parameter(torch.Tensor(1).fill_(init_scale)) 55 | self.weight = Variable(torch.Tensor(1).fill_(init_scale), requires_grad=False) 56 | self.weight = self.weight.cuda() 57 | if bias: 58 | self.bias = nn.Parameter(torch.Tensor(output_size).fill_(0)) 59 | self.proj = Variable(torch.Tensor( 60 | output_size, input_size), requires_grad=False) 61 | torch.manual_seed(123) 62 | nn.init.orthogonal(self.proj) 63 | 64 | def forward(self, x): 65 | w = self.proj.type_as(x) 66 | x = x / x.norm(2, -1, keepdim=True) 67 | out = nn.functional.linear(x, w) 68 | if hasattr(self, 'weight'): 69 | out = out * self.weight 70 | if hasattr(self, 'bias'): 71 | out = out + self.bias.view(1, -1) 72 | return out 73 | -------------------------------------------------------------------------------- /lib/model/fixed_proj.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/fixed_proj.pyc -------------------------------------------------------------------------------- /lib/model/nms/.gitignore: -------------------------------------------------------------------------------- 1 | *.c 2 | *.cpp 3 | *.so 4 | -------------------------------------------------------------------------------- /lib/model/nms/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/nms/__init__.py -------------------------------------------------------------------------------- /lib/model/nms/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/nms/__init__.pyc -------------------------------------------------------------------------------- /lib/model/nms/_ext/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/nms/_ext/__init__.py -------------------------------------------------------------------------------- /lib/model/nms/_ext/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/nms/_ext/__init__.pyc -------------------------------------------------------------------------------- /lib/model/nms/_ext/nms/__init__.py: -------------------------------------------------------------------------------- 1 | 2 | from torch.utils.ffi import _wrap_function 3 | from ._nms import lib as _lib, ffi as _ffi 4 | 5 | __all__ = [] 6 | def _import_symbols(locals): 7 | for symbol in dir(_lib): 8 | fn = getattr(_lib, symbol) 9 | if callable(fn): 10 | locals[symbol] = _wrap_function(fn, _ffi) 11 | else: 12 | locals[symbol] = fn 13 | __all__.append(symbol) 14 | 15 | _import_symbols(locals()) 16 | -------------------------------------------------------------------------------- /lib/model/nms/_ext/nms/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/nms/_ext/nms/__init__.pyc -------------------------------------------------------------------------------- /lib/model/nms/_ext/nms/_nms.so: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/nms/_ext/nms/_nms.so -------------------------------------------------------------------------------- /lib/model/nms/build.py: -------------------------------------------------------------------------------- 1 | from __future__ import print_function 2 | import os 3 | import torch 4 | from torch.utils.ffi import create_extension 5 | 6 | #this_file = os.path.dirname(__file__) 7 | 8 | sources = [] 9 | headers = [] 10 | defines = [] 11 | with_cuda = False 12 | 13 | if torch.cuda.is_available(): 14 | print('Including CUDA code.') 15 | sources += ['src/nms_cuda.c'] 16 | headers += ['src/nms_cuda.h'] 17 | defines += [('WITH_CUDA', None)] 18 | with_cuda = True 19 | 20 | this_file = os.path.dirname(os.path.realpath(__file__)) 21 | print(this_file) 22 | extra_objects = ['src/nms_cuda_kernel.cu.o'] 23 | extra_objects = [os.path.join(this_file, fname) for fname in extra_objects] 24 | print(extra_objects) 25 | 26 | ffi = create_extension( 27 | '_ext.nms', 28 | headers=headers, 29 | sources=sources, 30 | define_macros=defines, 31 | relative_to=__file__, 32 | with_cuda=with_cuda, 33 | extra_objects=extra_objects 34 | ) 35 | 36 | if __name__ == '__main__': 37 | ffi.build() 38 | -------------------------------------------------------------------------------- /lib/model/nms/make.sh: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env bash 2 | 3 | # CUDA_PATH=/usr/local/cuda/ 4 | 5 | cd src 6 | echo "Compiling stnm kernels by nvcc..." 7 | nvcc -c -o nms_cuda_kernel.cu.o nms_cuda_kernel.cu -x cu -Xcompiler -fPIC -arch=sm_52 8 | 9 | cd ../ 10 | python build.py 11 | -------------------------------------------------------------------------------- /lib/model/nms/nms_gpu.py: -------------------------------------------------------------------------------- 1 | from __future__ import absolute_import 2 | import torch 3 | import numpy as np 4 | from ._ext import nms 5 | import pdb 6 | 7 | def nms_gpu(dets, thresh): 8 | keep = dets.new(dets.size(0), 1).zero_().int() 9 | num_out = dets.new(1).zero_().int() 10 | nms.nms_cuda(keep, dets, num_out, thresh) 11 | keep = keep[:num_out[0]] 12 | return keep 13 | -------------------------------------------------------------------------------- /lib/model/nms/nms_gpu.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/nms/nms_gpu.pyc -------------------------------------------------------------------------------- /lib/model/nms/nms_kernel.cu: -------------------------------------------------------------------------------- 1 | // ------------------------------------------------------------------ 2 | // Faster R-CNN 3 | // Copyright (c) 2015 Microsoft 4 | // Licensed under The MIT License [see fast-rcnn/LICENSE for details] 5 | // Written by Shaoqing Ren 6 | // ------------------------------------------------------------------ 7 | 8 | #include "gpu_nms.hpp" 9 | #include 10 | #include 11 | 12 | #define CUDA_CHECK(condition) \ 13 | /* Code block avoids redefinition of cudaError_t error */ \ 14 | do { \ 15 | cudaError_t error = condition; \ 16 | if (error != cudaSuccess) { \ 17 | std::cout << cudaGetErrorString(error) << std::endl; \ 18 | } \ 19 | } while (0) 20 | 21 | #define DIVUP(m,n) ((m) / (n) + ((m) % (n) > 0)) 22 | int const threadsPerBlock = sizeof(unsigned long long) * 8; 23 | 24 | __device__ inline float devIoU(float const * const a, float const * const b) { 25 | float left = max(a[0], b[0]), right = min(a[2], b[2]); 26 | float top = max(a[1], b[1]), bottom = min(a[3], b[3]); 27 | float width = max(right - left + 1, 0.f), height = max(bottom - top + 1, 0.f); 28 | float interS = width * height; 29 | float Sa = (a[2] - a[0] + 1) * (a[3] - a[1] + 1); 30 | float Sb = (b[2] - b[0] + 1) * (b[3] - b[1] + 1); 31 | return interS / (Sa + Sb - interS); 32 | } 33 | 34 | __global__ void nms_kernel(const int n_boxes, const float nms_overlap_thresh, 35 | const float *dev_boxes, unsigned long long *dev_mask) { 36 | const int row_start = blockIdx.y; 37 | const int col_start = blockIdx.x; 38 | 39 | // if (row_start > col_start) return; 40 | 41 | const int row_size = 42 | min(n_boxes - row_start * threadsPerBlock, threadsPerBlock); 43 | const int col_size = 44 | min(n_boxes - col_start * threadsPerBlock, threadsPerBlock); 45 | 46 | __shared__ float block_boxes[threadsPerBlock * 5]; 47 | if (threadIdx.x < col_size) { 48 | block_boxes[threadIdx.x * 5 + 0] = 49 | dev_boxes[(threadsPerBlock * col_start + threadIdx.x) * 5 + 0]; 50 | block_boxes[threadIdx.x * 5 + 1] = 51 | dev_boxes[(threadsPerBlock * col_start + threadIdx.x) * 5 + 1]; 52 | block_boxes[threadIdx.x * 5 + 2] = 53 | dev_boxes[(threadsPerBlock * col_start + threadIdx.x) * 5 + 2]; 54 | block_boxes[threadIdx.x * 5 + 3] = 55 | dev_boxes[(threadsPerBlock * col_start + threadIdx.x) * 5 + 3]; 56 | block_boxes[threadIdx.x * 5 + 4] = 57 | dev_boxes[(threadsPerBlock * col_start + threadIdx.x) * 5 + 4]; 58 | } 59 | __syncthreads(); 60 | 61 | if (threadIdx.x < row_size) { 62 | const int cur_box_idx = threadsPerBlock * row_start + threadIdx.x; 63 | const float *cur_box = dev_boxes + cur_box_idx * 5; 64 | int i = 0; 65 | unsigned long long t = 0; 66 | int start = 0; 67 | if (row_start == col_start) { 68 | start = threadIdx.x + 1; 69 | } 70 | for (i = start; i < col_size; i++) { 71 | if (devIoU(cur_box, block_boxes + i * 5) > nms_overlap_thresh) { 72 | t |= 1ULL << i; 73 | } 74 | } 75 | const int col_blocks = DIVUP(n_boxes, threadsPerBlock); 76 | dev_mask[cur_box_idx * col_blocks + col_start] = t; 77 | } 78 | } 79 | 80 | void _set_device(int device_id) { 81 | int current_device; 82 | CUDA_CHECK(cudaGetDevice(¤t_device)); 83 | if (current_device == device_id) { 84 | return; 85 | } 86 | // The call to cudaSetDevice must come before any calls to Get, which 87 | // may perform initialization using the GPU. 88 | CUDA_CHECK(cudaSetDevice(device_id)); 89 | } 90 | 91 | void _nms(int* keep_out, int* num_out, const float* boxes_host, int boxes_num, 92 | int boxes_dim, float nms_overlap_thresh, int device_id) { 93 | _set_device(device_id); 94 | 95 | float* boxes_dev = NULL; 96 | unsigned long long* mask_dev = NULL; 97 | 98 | const int col_blocks = DIVUP(boxes_num, threadsPerBlock); 99 | 100 | CUDA_CHECK(cudaMalloc(&boxes_dev, 101 | boxes_num * boxes_dim * sizeof(float))); 102 | CUDA_CHECK(cudaMemcpy(boxes_dev, 103 | boxes_host, 104 | boxes_num * boxes_dim * sizeof(float), 105 | cudaMemcpyHostToDevice)); 106 | 107 | CUDA_CHECK(cudaMalloc(&mask_dev, 108 | boxes_num * col_blocks * sizeof(unsigned long long))); 109 | 110 | dim3 blocks(DIVUP(boxes_num, threadsPerBlock), 111 | DIVUP(boxes_num, threadsPerBlock)); 112 | dim3 threads(threadsPerBlock); 113 | nms_kernel<<>>(boxes_num, 114 | nms_overlap_thresh, 115 | boxes_dev, 116 | mask_dev); 117 | 118 | std::vector mask_host(boxes_num * col_blocks); 119 | CUDA_CHECK(cudaMemcpy(&mask_host[0], 120 | mask_dev, 121 | sizeof(unsigned long long) * boxes_num * col_blocks, 122 | cudaMemcpyDeviceToHost)); 123 | 124 | std::vector remv(col_blocks); 125 | memset(&remv[0], 0, sizeof(unsigned long long) * col_blocks); 126 | 127 | int num_to_keep = 0; 128 | for (int i = 0; i < boxes_num; i++) { 129 | int nblock = i / threadsPerBlock; 130 | int inblock = i % threadsPerBlock; 131 | 132 | if (!(remv[nblock] & (1ULL << inblock))) { 133 | keep_out[num_to_keep++] = i; 134 | unsigned long long *p = &mask_host[0] + i * col_blocks; 135 | for (int j = nblock; j < col_blocks; j++) { 136 | remv[j] |= p[j]; 137 | } 138 | } 139 | } 140 | *num_out = num_to_keep; 141 | 142 | CUDA_CHECK(cudaFree(boxes_dev)); 143 | CUDA_CHECK(cudaFree(mask_dev)); 144 | } 145 | -------------------------------------------------------------------------------- /lib/model/nms/nms_wrapper.py: -------------------------------------------------------------------------------- 1 | # -------------------------------------------------------- 2 | # Fast R-CNN 3 | # Copyright (c) 2015 Microsoft 4 | # Licensed under The MIT License [see LICENSE for details] 5 | # Written by Ross Girshick 6 | # -------------------------------------------------------- 7 | import torch 8 | from model.utils.config import cfg 9 | from model.nms.nms_gpu import nms_gpu 10 | 11 | def nms(dets, thresh, force_cpu=False): 12 | """Dispatch to either CPU or GPU NMS implementations.""" 13 | if dets.shape[0] == 0: 14 | return [] 15 | # ---numpy version--- 16 | # original: return gpu_nms(dets, thresh, device_id=cfg.GPU_ID) 17 | # ---pytorch version--- 18 | return nms_gpu(dets, thresh) 19 | -------------------------------------------------------------------------------- /lib/model/nms/nms_wrapper.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/nms/nms_wrapper.pyc -------------------------------------------------------------------------------- /lib/model/nms/src/nms_cuda.c: -------------------------------------------------------------------------------- 1 | #include 2 | #include 3 | #include "nms_cuda_kernel.h" 4 | 5 | // this symbol will be resolved automatically from PyTorch libs 6 | extern THCState *state; 7 | 8 | int nms_cuda(THCudaIntTensor *keep_out, THCudaTensor *boxes_host, 9 | THCudaIntTensor *num_out, float nms_overlap_thresh) { 10 | 11 | nms_cuda_compute(THCudaIntTensor_data(state, keep_out), 12 | THCudaIntTensor_data(state, num_out), 13 | THCudaTensor_data(state, boxes_host), 14 | boxes_host->size[0], 15 | boxes_host->size[1], 16 | nms_overlap_thresh); 17 | 18 | return 1; 19 | } 20 | -------------------------------------------------------------------------------- /lib/model/nms/src/nms_cuda.h: -------------------------------------------------------------------------------- 1 | // int nms_cuda(THCudaTensor *keep_out, THCudaTensor *num_out, 2 | // THCudaTensor *boxes_host, THCudaTensor *nms_overlap_thresh); 3 | 4 | int nms_cuda(THCudaIntTensor *keep_out, THCudaTensor *boxes_host, 5 | THCudaIntTensor *num_out, float nms_overlap_thresh); 6 | -------------------------------------------------------------------------------- /lib/model/nms/src/nms_cuda_kernel.cu: -------------------------------------------------------------------------------- 1 | // ------------------------------------------------------------------ 2 | // Faster R-CNN 3 | // Copyright (c) 2015 Microsoft 4 | // Licensed under The MIT License [see fast-rcnn/LICENSE for details] 5 | // Written by Shaoqing Ren 6 | // ------------------------------------------------------------------ 7 | 8 | #include 9 | #include 10 | #include 11 | #include 12 | #include "nms_cuda_kernel.h" 13 | 14 | #define CUDA_WARN(XXX) \ 15 | do { if (XXX != cudaSuccess) std::cout << "CUDA Error: " << \ 16 | cudaGetErrorString(XXX) << ", at line " << __LINE__ \ 17 | << std::endl; cudaDeviceSynchronize(); } while (0) 18 | 19 | #define CUDA_CHECK(condition) \ 20 | /* Code block avoids redefinition of cudaError_t error */ \ 21 | do { \ 22 | cudaError_t error = condition; \ 23 | if (error != cudaSuccess) { \ 24 | std::cout << cudaGetErrorString(error) << std::endl; \ 25 | } \ 26 | } while (0) 27 | 28 | #define DIVUP(m,n) ((m) / (n) + ((m) % (n) > 0)) 29 | int const threadsPerBlock = sizeof(unsigned long long) * 8; 30 | 31 | __device__ inline float devIoU(float const * const a, float const * const b) { 32 | float left = max(a[0], b[0]), right = min(a[2], b[2]); 33 | float top = max(a[1], b[1]), bottom = min(a[3], b[3]); 34 | float width = max(right - left + 1, 0.f), height = max(bottom - top + 1, 0.f); 35 | float interS = width * height; 36 | float Sa = (a[2] - a[0] + 1) * (a[3] - a[1] + 1); 37 | float Sb = (b[2] - b[0] + 1) * (b[3] - b[1] + 1); 38 | return interS / (Sa + Sb - interS); 39 | } 40 | 41 | __global__ void nms_kernel(int n_boxes, float nms_overlap_thresh, 42 | float *dev_boxes, unsigned long long *dev_mask) { 43 | const int row_start = blockIdx.y; 44 | const int col_start = blockIdx.x; 45 | 46 | // if (row_start > col_start) return; 47 | 48 | const int row_size = 49 | min(n_boxes - row_start * threadsPerBlock, threadsPerBlock); 50 | const int col_size = 51 | min(n_boxes - col_start * threadsPerBlock, threadsPerBlock); 52 | 53 | __shared__ float block_boxes[threadsPerBlock * 5]; 54 | if (threadIdx.x < col_size) { 55 | block_boxes[threadIdx.x * 5 + 0] = 56 | dev_boxes[(threadsPerBlock * col_start + threadIdx.x) * 5 + 0]; 57 | block_boxes[threadIdx.x * 5 + 1] = 58 | dev_boxes[(threadsPerBlock * col_start + threadIdx.x) * 5 + 1]; 59 | block_boxes[threadIdx.x * 5 + 2] = 60 | dev_boxes[(threadsPerBlock * col_start + threadIdx.x) * 5 + 2]; 61 | block_boxes[threadIdx.x * 5 + 3] = 62 | dev_boxes[(threadsPerBlock * col_start + threadIdx.x) * 5 + 3]; 63 | block_boxes[threadIdx.x * 5 + 4] = 64 | dev_boxes[(threadsPerBlock * col_start + threadIdx.x) * 5 + 4]; 65 | } 66 | __syncthreads(); 67 | 68 | if (threadIdx.x < row_size) { 69 | const int cur_box_idx = threadsPerBlock * row_start + threadIdx.x; 70 | const float *cur_box = dev_boxes + cur_box_idx * 5; 71 | int i = 0; 72 | unsigned long long t = 0; 73 | int start = 0; 74 | if (row_start == col_start) { 75 | start = threadIdx.x + 1; 76 | } 77 | for (i = start; i < col_size; i++) { 78 | if (devIoU(cur_box, block_boxes + i * 5) > nms_overlap_thresh) { 79 | t |= 1ULL << i; 80 | } 81 | } 82 | const int col_blocks = DIVUP(n_boxes, threadsPerBlock); 83 | dev_mask[cur_box_idx * col_blocks + col_start] = t; 84 | } 85 | } 86 | 87 | void nms_cuda_compute(int* keep_out, int *num_out, float* boxes_host, int boxes_num, 88 | int boxes_dim, float nms_overlap_thresh) { 89 | 90 | float* boxes_dev = NULL; 91 | unsigned long long* mask_dev = NULL; 92 | 93 | const int col_blocks = DIVUP(boxes_num, threadsPerBlock); 94 | 95 | CUDA_CHECK(cudaMalloc(&boxes_dev, 96 | boxes_num * boxes_dim * sizeof(float))); 97 | CUDA_CHECK(cudaMemcpy(boxes_dev, 98 | boxes_host, 99 | boxes_num * boxes_dim * sizeof(float), 100 | cudaMemcpyHostToDevice)); 101 | 102 | CUDA_CHECK(cudaMalloc(&mask_dev, 103 | boxes_num * col_blocks * sizeof(unsigned long long))); 104 | 105 | dim3 blocks(DIVUP(boxes_num, threadsPerBlock), 106 | DIVUP(boxes_num, threadsPerBlock)); 107 | dim3 threads(threadsPerBlock); 108 | 109 | // printf("i am at line %d\n", boxes_num); 110 | // printf("i am at line %d\n", boxes_dim); 111 | 112 | nms_kernel<<>>(boxes_num, 113 | nms_overlap_thresh, 114 | boxes_dev, 115 | mask_dev); 116 | 117 | std::vector mask_host(boxes_num * col_blocks); 118 | CUDA_CHECK(cudaMemcpy(&mask_host[0], 119 | mask_dev, 120 | sizeof(unsigned long long) * boxes_num * col_blocks, 121 | cudaMemcpyDeviceToHost)); 122 | 123 | std::vector remv(col_blocks); 124 | memset(&remv[0], 0, sizeof(unsigned long long) * col_blocks); 125 | 126 | // we need to create a memory for keep_out on cpu 127 | // otherwise, the following code cannot run 128 | 129 | int* keep_out_cpu = new int[boxes_num]; 130 | 131 | int num_to_keep = 0; 132 | for (int i = 0; i < boxes_num; i++) { 133 | int nblock = i / threadsPerBlock; 134 | int inblock = i % threadsPerBlock; 135 | 136 | if (!(remv[nblock] & (1ULL << inblock))) { 137 | // orignal: keep_out[num_to_keep++] = i; 138 | keep_out_cpu[num_to_keep++] = i; 139 | unsigned long long *p = &mask_host[0] + i * col_blocks; 140 | for (int j = nblock; j < col_blocks; j++) { 141 | remv[j] |= p[j]; 142 | } 143 | } 144 | } 145 | 146 | // copy keep_out_cpu to keep_out on gpu 147 | CUDA_WARN(cudaMemcpy(keep_out, keep_out_cpu, boxes_num * sizeof(int),cudaMemcpyHostToDevice)); 148 | 149 | // *num_out = num_to_keep; 150 | 151 | // original: *num_out = num_to_keep; 152 | // copy num_to_keep to num_out on gpu 153 | 154 | CUDA_WARN(cudaMemcpy(num_out, &num_to_keep, 1 * sizeof(int),cudaMemcpyHostToDevice)); 155 | 156 | // release cuda memory 157 | CUDA_CHECK(cudaFree(boxes_dev)); 158 | CUDA_CHECK(cudaFree(mask_dev)); 159 | // release cpu memory 160 | delete []keep_out_cpu; 161 | } 162 | -------------------------------------------------------------------------------- /lib/model/nms/src/nms_cuda_kernel.cu.o: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/nms/src/nms_cuda_kernel.cu.o -------------------------------------------------------------------------------- /lib/model/nms/src/nms_cuda_kernel.h: -------------------------------------------------------------------------------- 1 | #ifdef __cplusplus 2 | extern "C" { 3 | #endif 4 | 5 | void nms_cuda_compute(int* keep_out, int *num_out, float* boxes_host, int boxes_num, 6 | int boxes_dim, float nms_overlap_thresh); 7 | 8 | #ifdef __cplusplus 9 | } 10 | #endif 11 | -------------------------------------------------------------------------------- /lib/model/roi_align/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_align/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_align/_ext/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/_ext/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_align/_ext/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/_ext/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_align/_ext/roi_align/__init__.py: -------------------------------------------------------------------------------- 1 | 2 | from torch.utils.ffi import _wrap_function 3 | from ._roi_align import lib as _lib, ffi as _ffi 4 | 5 | __all__ = [] 6 | def _import_symbols(locals): 7 | for symbol in dir(_lib): 8 | fn = getattr(_lib, symbol) 9 | if callable(fn): 10 | locals[symbol] = _wrap_function(fn, _ffi) 11 | else: 12 | locals[symbol] = fn 13 | __all__.append(symbol) 14 | 15 | _import_symbols(locals()) 16 | -------------------------------------------------------------------------------- /lib/model/roi_align/_ext/roi_align/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/_ext/roi_align/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_align/_ext/roi_align/_roi_align.so: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/_ext/roi_align/_roi_align.so -------------------------------------------------------------------------------- /lib/model/roi_align/build.py: -------------------------------------------------------------------------------- 1 | from __future__ import print_function 2 | import os 3 | import torch 4 | from torch.utils.ffi import create_extension 5 | 6 | # sources = ['src/roi_align.c'] 7 | # headers = ['src/roi_align.h'] 8 | sources = [] 9 | headers = [] 10 | defines = [] 11 | with_cuda = False 12 | 13 | if torch.cuda.is_available(): 14 | print('Including CUDA code.') 15 | sources += ['src/roi_align_cuda.c'] 16 | headers += ['src/roi_align_cuda.h'] 17 | defines += [('WITH_CUDA', None)] 18 | with_cuda = True 19 | 20 | this_file = os.path.dirname(os.path.realpath(__file__)) 21 | print(this_file) 22 | extra_objects = ['src/roi_align_kernel.cu.o'] 23 | extra_objects = [os.path.join(this_file, fname) for fname in extra_objects] 24 | 25 | ffi = create_extension( 26 | '_ext.roi_align', 27 | headers=headers, 28 | sources=sources, 29 | define_macros=defines, 30 | relative_to=__file__, 31 | with_cuda=with_cuda, 32 | extra_objects=extra_objects 33 | ) 34 | 35 | if __name__ == '__main__': 36 | ffi.build() 37 | -------------------------------------------------------------------------------- /lib/model/roi_align/functions/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/functions/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_align/functions/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/functions/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_align/functions/roi_align.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch.autograd import Function 3 | from .._ext import roi_align 4 | 5 | 6 | # TODO use save_for_backward instead 7 | class RoIAlignFunction(Function): 8 | def __init__(self, aligned_height, aligned_width, spatial_scale): 9 | self.aligned_width = int(aligned_width) 10 | self.aligned_height = int(aligned_height) 11 | self.spatial_scale = float(spatial_scale) 12 | self.rois = None 13 | self.feature_size = None 14 | 15 | def forward(self, features, rois): 16 | self.rois = rois 17 | self.feature_size = features.size() 18 | 19 | batch_size, num_channels, data_height, data_width = features.size() 20 | num_rois = rois.size(0) 21 | 22 | output = features.new(num_rois, num_channels, self.aligned_height, self.aligned_width).zero_() 23 | if features.is_cuda: 24 | roi_align.roi_align_forward_cuda(self.aligned_height, 25 | self.aligned_width, 26 | self.spatial_scale, features, 27 | rois, output) 28 | else: 29 | raise NotImplementedError 30 | 31 | return output 32 | 33 | def backward(self, grad_output): 34 | assert(self.feature_size is not None and grad_output.is_cuda) 35 | 36 | batch_size, num_channels, data_height, data_width = self.feature_size 37 | 38 | grad_input = self.rois.new(batch_size, num_channels, data_height, 39 | data_width).zero_() 40 | roi_align.roi_align_backward_cuda(self.aligned_height, 41 | self.aligned_width, 42 | self.spatial_scale, grad_output, 43 | self.rois, grad_input) 44 | 45 | # print grad_input 46 | 47 | return grad_input, None 48 | -------------------------------------------------------------------------------- /lib/model/roi_align/functions/roi_align.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/functions/roi_align.pyc -------------------------------------------------------------------------------- /lib/model/roi_align/make.sh: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env bash 2 | 3 | CUDA_PATH=/usr/local/cuda/ 4 | 5 | cd src 6 | echo "Compiling my_lib kernels by nvcc..." 7 | nvcc -c -o roi_align_kernel.cu.o roi_align_kernel.cu -x cu -Xcompiler -fPIC -arch=sm_52 8 | 9 | cd ../ 10 | python build.py 11 | -------------------------------------------------------------------------------- /lib/model/roi_align/modules/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/modules/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_align/modules/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/modules/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_align/modules/roi_align.py: -------------------------------------------------------------------------------- 1 | from torch.nn.modules.module import Module 2 | from torch.nn.functional import avg_pool2d, max_pool2d 3 | from ..functions.roi_align import RoIAlignFunction 4 | 5 | 6 | class RoIAlign(Module): 7 | def __init__(self, aligned_height, aligned_width, spatial_scale): 8 | super(RoIAlign, self).__init__() 9 | 10 | self.aligned_width = int(aligned_width) 11 | self.aligned_height = int(aligned_height) 12 | self.spatial_scale = float(spatial_scale) 13 | 14 | def forward(self, features, rois): 15 | return RoIAlignFunction(self.aligned_height, self.aligned_width, 16 | self.spatial_scale)(features, rois) 17 | 18 | class RoIAlignAvg(Module): 19 | def __init__(self, aligned_height, aligned_width, spatial_scale): 20 | super(RoIAlignAvg, self).__init__() 21 | 22 | self.aligned_width = int(aligned_width) 23 | self.aligned_height = int(aligned_height) 24 | self.spatial_scale = float(spatial_scale) 25 | 26 | def forward(self, features, rois): 27 | x = RoIAlignFunction(self.aligned_height+1, self.aligned_width+1, 28 | self.spatial_scale)(features, rois) 29 | return avg_pool2d(x, kernel_size=2, stride=1) 30 | 31 | class RoIAlignMax(Module): 32 | def __init__(self, aligned_height, aligned_width, spatial_scale): 33 | super(RoIAlignMax, self).__init__() 34 | 35 | self.aligned_width = int(aligned_width) 36 | self.aligned_height = int(aligned_height) 37 | self.spatial_scale = float(spatial_scale) 38 | 39 | def forward(self, features, rois): 40 | x = RoIAlignFunction(self.aligned_height+1, self.aligned_width+1, 41 | self.spatial_scale)(features, rois) 42 | return max_pool2d(x, kernel_size=2, stride=1) 43 | -------------------------------------------------------------------------------- /lib/model/roi_align/modules/roi_align.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/modules/roi_align.pyc -------------------------------------------------------------------------------- /lib/model/roi_align/src/roi_align_cuda.c: -------------------------------------------------------------------------------- 1 | #include 2 | #include 3 | #include "roi_align_kernel.h" 4 | 5 | extern THCState *state; 6 | 7 | int roi_align_forward_cuda(int aligned_height, int aligned_width, float spatial_scale, 8 | THCudaTensor * features, THCudaTensor * rois, THCudaTensor * output) 9 | { 10 | // Grab the input tensor 11 | float * data_flat = THCudaTensor_data(state, features); 12 | float * rois_flat = THCudaTensor_data(state, rois); 13 | 14 | float * output_flat = THCudaTensor_data(state, output); 15 | 16 | // Number of ROIs 17 | int num_rois = THCudaTensor_size(state, rois, 0); 18 | int size_rois = THCudaTensor_size(state, rois, 1); 19 | if (size_rois != 5) 20 | { 21 | return 0; 22 | } 23 | 24 | // data height 25 | int data_height = THCudaTensor_size(state, features, 2); 26 | // data width 27 | int data_width = THCudaTensor_size(state, features, 3); 28 | // Number of channels 29 | int num_channels = THCudaTensor_size(state, features, 1); 30 | 31 | cudaStream_t stream = THCState_getCurrentStream(state); 32 | 33 | ROIAlignForwardLaucher( 34 | data_flat, spatial_scale, num_rois, data_height, 35 | data_width, num_channels, aligned_height, 36 | aligned_width, rois_flat, 37 | output_flat, stream); 38 | 39 | return 1; 40 | } 41 | 42 | int roi_align_backward_cuda(int aligned_height, int aligned_width, float spatial_scale, 43 | THCudaTensor * top_grad, THCudaTensor * rois, THCudaTensor * bottom_grad) 44 | { 45 | // Grab the input tensor 46 | float * top_grad_flat = THCudaTensor_data(state, top_grad); 47 | float * rois_flat = THCudaTensor_data(state, rois); 48 | 49 | float * bottom_grad_flat = THCudaTensor_data(state, bottom_grad); 50 | 51 | // Number of ROIs 52 | int num_rois = THCudaTensor_size(state, rois, 0); 53 | int size_rois = THCudaTensor_size(state, rois, 1); 54 | if (size_rois != 5) 55 | { 56 | return 0; 57 | } 58 | 59 | // batch size 60 | int batch_size = THCudaTensor_size(state, bottom_grad, 0); 61 | // data height 62 | int data_height = THCudaTensor_size(state, bottom_grad, 2); 63 | // data width 64 | int data_width = THCudaTensor_size(state, bottom_grad, 3); 65 | // Number of channels 66 | int num_channels = THCudaTensor_size(state, bottom_grad, 1); 67 | 68 | cudaStream_t stream = THCState_getCurrentStream(state); 69 | ROIAlignBackwardLaucher( 70 | top_grad_flat, spatial_scale, batch_size, num_rois, data_height, 71 | data_width, num_channels, aligned_height, 72 | aligned_width, rois_flat, 73 | bottom_grad_flat, stream); 74 | 75 | return 1; 76 | } 77 | -------------------------------------------------------------------------------- /lib/model/roi_align/src/roi_align_cuda.h: -------------------------------------------------------------------------------- 1 | int roi_align_forward_cuda(int aligned_height, int aligned_width, float spatial_scale, 2 | THCudaTensor * features, THCudaTensor * rois, THCudaTensor * output); 3 | 4 | int roi_align_backward_cuda(int aligned_height, int aligned_width, float spatial_scale, 5 | THCudaTensor * top_grad, THCudaTensor * rois, THCudaTensor * bottom_grad); 6 | -------------------------------------------------------------------------------- /lib/model/roi_align/src/roi_align_kernel.cu.o: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_align/src/roi_align_kernel.cu.o -------------------------------------------------------------------------------- /lib/model/roi_align/src/roi_align_kernel.h: -------------------------------------------------------------------------------- 1 | #ifndef _ROI_ALIGN_KERNEL 2 | #define _ROI_ALIGN_KERNEL 3 | 4 | #ifdef __cplusplus 5 | extern "C" { 6 | #endif 7 | 8 | __global__ void ROIAlignForward(const int nthreads, const float* bottom_data, 9 | const float spatial_scale, const int height, const int width, 10 | const int channels, const int aligned_height, const int aligned_width, 11 | const float* bottom_rois, float* top_data); 12 | 13 | int ROIAlignForwardLaucher( 14 | const float* bottom_data, const float spatial_scale, const int num_rois, const int height, 15 | const int width, const int channels, const int aligned_height, 16 | const int aligned_width, const float* bottom_rois, 17 | float* top_data, cudaStream_t stream); 18 | 19 | __global__ void ROIAlignBackward(const int nthreads, const float* top_diff, 20 | const float spatial_scale, const int height, const int width, 21 | const int channels, const int aligned_height, const int aligned_width, 22 | float* bottom_diff, const float* bottom_rois); 23 | 24 | int ROIAlignBackwardLaucher(const float* top_diff, const float spatial_scale, const int batch_size, const int num_rois, 25 | const int height, const int width, const int channels, const int aligned_height, 26 | const int aligned_width, const float* bottom_rois, 27 | float* bottom_diff, cudaStream_t stream); 28 | 29 | #ifdef __cplusplus 30 | } 31 | #endif 32 | 33 | #endif 34 | 35 | -------------------------------------------------------------------------------- /lib/model/roi_crop/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_crop/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_crop/_ext/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/_ext/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_crop/_ext/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/_ext/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_crop/_ext/crop_resize/__init__.py: -------------------------------------------------------------------------------- 1 | 2 | from torch.utils.ffi import _wrap_function 3 | from ._crop_resize import lib as _lib, ffi as _ffi 4 | 5 | __all__ = [] 6 | def _import_symbols(locals): 7 | for symbol in dir(_lib): 8 | fn = getattr(_lib, symbol) 9 | locals[symbol] = _wrap_function(fn, _ffi) 10 | __all__.append(symbol) 11 | 12 | _import_symbols(locals()) 13 | -------------------------------------------------------------------------------- /lib/model/roi_crop/_ext/crop_resize/_crop_resize.so: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/_ext/crop_resize/_crop_resize.so -------------------------------------------------------------------------------- /lib/model/roi_crop/_ext/roi_crop/__init__.py: -------------------------------------------------------------------------------- 1 | 2 | from torch.utils.ffi import _wrap_function 3 | from ._roi_crop import lib as _lib, ffi as _ffi 4 | 5 | __all__ = [] 6 | def _import_symbols(locals): 7 | for symbol in dir(_lib): 8 | fn = getattr(_lib, symbol) 9 | if callable(fn): 10 | locals[symbol] = _wrap_function(fn, _ffi) 11 | else: 12 | locals[symbol] = fn 13 | __all__.append(symbol) 14 | 15 | _import_symbols(locals()) 16 | -------------------------------------------------------------------------------- /lib/model/roi_crop/_ext/roi_crop/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/_ext/roi_crop/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_crop/_ext/roi_crop/_roi_crop.so: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/_ext/roi_crop/_roi_crop.so -------------------------------------------------------------------------------- /lib/model/roi_crop/build.py: -------------------------------------------------------------------------------- 1 | from __future__ import print_function 2 | import os 3 | import torch 4 | from torch.utils.ffi import create_extension 5 | 6 | #this_file = os.path.dirname(__file__) 7 | 8 | sources = ['src/roi_crop.c'] 9 | headers = ['src/roi_crop.h'] 10 | defines = [] 11 | with_cuda = False 12 | 13 | if torch.cuda.is_available(): 14 | print('Including CUDA code.') 15 | sources += ['src/roi_crop_cuda.c'] 16 | headers += ['src/roi_crop_cuda.h'] 17 | defines += [('WITH_CUDA', None)] 18 | with_cuda = True 19 | 20 | this_file = os.path.dirname(os.path.realpath(__file__)) 21 | print(this_file) 22 | extra_objects = ['src/roi_crop_cuda_kernel.cu.o'] 23 | extra_objects = [os.path.join(this_file, fname) for fname in extra_objects] 24 | 25 | ffi = create_extension( 26 | '_ext.roi_crop', 27 | headers=headers, 28 | sources=sources, 29 | define_macros=defines, 30 | relative_to=__file__, 31 | with_cuda=with_cuda, 32 | extra_objects=extra_objects 33 | ) 34 | 35 | if __name__ == '__main__': 36 | ffi.build() 37 | -------------------------------------------------------------------------------- /lib/model/roi_crop/functions/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/functions/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_crop/functions/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/functions/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_crop/functions/crop_resize.py: -------------------------------------------------------------------------------- 1 | # functions/add.py 2 | import torch 3 | from torch.autograd import Function 4 | from .._ext import roi_crop 5 | from cffi import FFI 6 | ffi = FFI() 7 | 8 | class RoICropFunction(Function): 9 | def forward(self, input1, input2): 10 | self.input1 = input1 11 | self.input2 = input2 12 | self.device_c = ffi.new("int *") 13 | output = torch.zeros(input2.size()[0], input1.size()[1], input2.size()[1], input2.size()[2]) 14 | #print('decice %d' % torch.cuda.current_device()) 15 | if input1.is_cuda: 16 | self.device = torch.cuda.current_device() 17 | else: 18 | self.device = -1 19 | self.device_c[0] = self.device 20 | if not input1.is_cuda: 21 | roi_crop.BilinearSamplerBHWD_updateOutput(input1, input2, output) 22 | else: 23 | output = output.cuda(self.device) 24 | roi_crop.BilinearSamplerBHWD_updateOutput_cuda(input1, input2, output) 25 | return output 26 | 27 | def backward(self, grad_output): 28 | grad_input1 = torch.zeros(self.input1.size()) 29 | grad_input2 = torch.zeros(self.input2.size()) 30 | #print('backward decice %d' % self.device) 31 | if not grad_output.is_cuda: 32 | roi_crop.BilinearSamplerBHWD_updateGradInput(self.input1, self.input2, grad_input1, grad_input2, grad_output) 33 | else: 34 | grad_input1 = grad_input1.cuda(self.device) 35 | grad_input2 = grad_input2.cuda(self.device) 36 | roi_crop.BilinearSamplerBHWD_updateGradInput_cuda(self.input1, self.input2, grad_input1, grad_input2, grad_output) 37 | return grad_input1, grad_input2 38 | -------------------------------------------------------------------------------- /lib/model/roi_crop/functions/gridgen.py: -------------------------------------------------------------------------------- 1 | # functions/add.py 2 | import torch 3 | from torch.autograd import Function 4 | import numpy as np 5 | 6 | 7 | class AffineGridGenFunction(Function): 8 | def __init__(self, height, width,lr=1): 9 | super(AffineGridGenFunction, self).__init__() 10 | self.lr = lr 11 | self.height, self.width = height, width 12 | self.grid = np.zeros( [self.height, self.width, 3], dtype=np.float32) 13 | self.grid[:,:,0] = np.expand_dims(np.repeat(np.expand_dims(np.arange(-1, 1, 2.0/(self.height)), 0), repeats = self.width, axis = 0).T, 0) 14 | self.grid[:,:,1] = np.expand_dims(np.repeat(np.expand_dims(np.arange(-1, 1, 2.0/(self.width)), 0), repeats = self.height, axis = 0), 0) 15 | # self.grid[:,:,0] = np.expand_dims(np.repeat(np.expand_dims(np.arange(-1, 1, 2.0/(self.height - 1)), 0), repeats = self.width, axis = 0).T, 0) 16 | # self.grid[:,:,1] = np.expand_dims(np.repeat(np.expand_dims(np.arange(-1, 1, 2.0/(self.width - 1)), 0), repeats = self.height, axis = 0), 0) 17 | self.grid[:,:,2] = np.ones([self.height, width]) 18 | self.grid = torch.from_numpy(self.grid.astype(np.float32)) 19 | #print(self.grid) 20 | 21 | def forward(self, input1): 22 | self.input1 = input1 23 | output = input1.new(torch.Size([input1.size(0)]) + self.grid.size()).zero_() 24 | self.batchgrid = input1.new(torch.Size([input1.size(0)]) + self.grid.size()).zero_() 25 | for i in range(input1.size(0)): 26 | self.batchgrid[i] = self.grid.astype(self.batchgrid[i]) 27 | 28 | # if input1.is_cuda: 29 | # self.batchgrid = self.batchgrid.cuda() 30 | # output = output.cuda() 31 | 32 | for i in range(input1.size(0)): 33 | output = torch.bmm(self.batchgrid.view(-1, self.height*self.width, 3), torch.transpose(input1, 1, 2)).view(-1, self.height, self.width, 2) 34 | 35 | return output 36 | 37 | def backward(self, grad_output): 38 | 39 | grad_input1 = self.input1.new(self.input1.size()).zero_() 40 | 41 | # if grad_output.is_cuda: 42 | # self.batchgrid = self.batchgrid.cuda() 43 | # grad_input1 = grad_input1.cuda() 44 | 45 | grad_input1 = torch.baddbmm(grad_input1, torch.transpose(grad_output.view(-1, self.height*self.width, 2), 1,2), self.batchgrid.view(-1, self.height*self.width, 3)) 46 | return grad_input1 47 | -------------------------------------------------------------------------------- /lib/model/roi_crop/functions/roi_crop.py: -------------------------------------------------------------------------------- 1 | # functions/add.py 2 | import torch 3 | from torch.autograd import Function 4 | from .._ext import roi_crop 5 | import pdb 6 | 7 | class RoICropFunction(Function): 8 | def forward(self, input1, input2): 9 | self.input1 = input1.clone() 10 | self.input2 = input2.clone() 11 | output = input2.new(input2.size()[0], input1.size()[1], input2.size()[1], input2.size()[2]).zero_() 12 | assert output.get_device() == input1.get_device(), "output and input1 must on the same device" 13 | assert output.get_device() == input2.get_device(), "output and input2 must on the same device" 14 | roi_crop.BilinearSamplerBHWD_updateOutput_cuda(input1, input2, output) 15 | return output 16 | 17 | def backward(self, grad_output): 18 | grad_input1 = self.input1.new(self.input1.size()).zero_() 19 | grad_input2 = self.input2.new(self.input2.size()).zero_() 20 | roi_crop.BilinearSamplerBHWD_updateGradInput_cuda(self.input1, self.input2, grad_input1, grad_input2, grad_output) 21 | return grad_input1, grad_input2 22 | -------------------------------------------------------------------------------- /lib/model/roi_crop/functions/roi_crop.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/functions/roi_crop.pyc -------------------------------------------------------------------------------- /lib/model/roi_crop/make.sh: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env bash 2 | 3 | CUDA_PATH=/usr/local/cuda/ 4 | 5 | cd src 6 | echo "Compiling my_lib kernels by nvcc..." 7 | nvcc -c -o roi_crop_cuda_kernel.cu.o roi_crop_cuda_kernel.cu -x cu -Xcompiler -fPIC -arch=sm_52 8 | 9 | cd ../ 10 | python build.py 11 | -------------------------------------------------------------------------------- /lib/model/roi_crop/modules/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/modules/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_crop/modules/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/modules/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_crop/modules/roi_crop.py: -------------------------------------------------------------------------------- 1 | from torch.nn.modules.module import Module 2 | from ..functions.roi_crop import RoICropFunction 3 | 4 | class _RoICrop(Module): 5 | def __init__(self, layout = 'BHWD'): 6 | super(_RoICrop, self).__init__() 7 | def forward(self, input1, input2): 8 | return RoICropFunction()(input1, input2) 9 | -------------------------------------------------------------------------------- /lib/model/roi_crop/modules/roi_crop.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/modules/roi_crop.pyc -------------------------------------------------------------------------------- /lib/model/roi_crop/src/roi_crop.h: -------------------------------------------------------------------------------- 1 | int BilinearSamplerBHWD_updateOutput(THFloatTensor *inputImages, THFloatTensor *grids, THFloatTensor *output); 2 | 3 | int BilinearSamplerBHWD_updateGradInput(THFloatTensor *inputImages, THFloatTensor *grids, THFloatTensor *gradInputImages, 4 | THFloatTensor *gradGrids, THFloatTensor *gradOutput); 5 | 6 | 7 | 8 | int BilinearSamplerBCHW_updateOutput(THFloatTensor *inputImages, THFloatTensor *grids, THFloatTensor *output); 9 | 10 | int BilinearSamplerBCHW_updateGradInput(THFloatTensor *inputImages, THFloatTensor *grids, THFloatTensor *gradInputImages, 11 | THFloatTensor *gradGrids, THFloatTensor *gradOutput); 12 | -------------------------------------------------------------------------------- /lib/model/roi_crop/src/roi_crop_cuda.c: -------------------------------------------------------------------------------- 1 | #include 2 | #include 3 | #include 4 | #include "roi_crop_cuda_kernel.h" 5 | 6 | #define real float 7 | 8 | // this symbol will be resolved automatically from PyTorch libs 9 | extern THCState *state; 10 | 11 | // Bilinear sampling is done in BHWD (coalescing is not obvious in BDHW) 12 | // we assume BHWD format in inputImages 13 | // we assume BHW(YX) format on grids 14 | 15 | int BilinearSamplerBHWD_updateOutput_cuda(THCudaTensor *inputImages, THCudaTensor *grids, THCudaTensor *output){ 16 | // THCState *state = getCutorchState(L); 17 | // THCudaTensor *inputImages = (THCudaTensor *)luaT_checkudata(L, 2, "torch.CudaTensor"); 18 | // THCudaTensor *grids = (THCudaTensor *)luaT_checkudata(L, 3, "torch.CudaTensor"); 19 | // THCudaTensor *output = (THCudaTensor *)luaT_checkudata(L, 4, "torch.CudaTensor"); 20 | 21 | int success = 0; 22 | success = BilinearSamplerBHWD_updateOutput_cuda_kernel(output->size[1], 23 | output->size[3], 24 | output->size[2], 25 | output->size[0], 26 | THCudaTensor_size(state, inputImages, 1), 27 | THCudaTensor_size(state, inputImages, 2), 28 | THCudaTensor_size(state, inputImages, 3), 29 | THCudaTensor_size(state, inputImages, 0), 30 | THCudaTensor_data(state, inputImages), 31 | THCudaTensor_stride(state, inputImages, 0), 32 | THCudaTensor_stride(state, inputImages, 1), 33 | THCudaTensor_stride(state, inputImages, 2), 34 | THCudaTensor_stride(state, inputImages, 3), 35 | THCudaTensor_data(state, grids), 36 | THCudaTensor_stride(state, grids, 0), 37 | THCudaTensor_stride(state, grids, 3), 38 | THCudaTensor_stride(state, grids, 1), 39 | THCudaTensor_stride(state, grids, 2), 40 | THCudaTensor_data(state, output), 41 | THCudaTensor_stride(state, output, 0), 42 | THCudaTensor_stride(state, output, 1), 43 | THCudaTensor_stride(state, output, 2), 44 | THCudaTensor_stride(state, output, 3), 45 | THCState_getCurrentStream(state)); 46 | 47 | //check for errors 48 | if (!success) { 49 | THError("aborting"); 50 | } 51 | return 1; 52 | } 53 | 54 | int BilinearSamplerBHWD_updateGradInput_cuda(THCudaTensor *inputImages, THCudaTensor *grids, THCudaTensor *gradInputImages, 55 | THCudaTensor *gradGrids, THCudaTensor *gradOutput) 56 | { 57 | // THCState *state = getCutorchState(L); 58 | // THCudaTensor *inputImages = (THCudaTensor *)luaT_checkudata(L, 2, "torch.CudaTensor"); 59 | // THCudaTensor *grids = (THCudaTensor *)luaT_checkudata(L, 3, "torch.CudaTensor"); 60 | // THCudaTensor *gradInputImages = (THCudaTensor *)luaT_checkudata(L, 4, "torch.CudaTensor"); 61 | // THCudaTensor *gradGrids = (THCudaTensor *)luaT_checkudata(L, 5, "torch.CudaTensor"); 62 | // THCudaTensor *gradOutput = (THCudaTensor *)luaT_checkudata(L, 6, "torch.CudaTensor"); 63 | 64 | int success = 0; 65 | success = BilinearSamplerBHWD_updateGradInput_cuda_kernel(gradOutput->size[1], 66 | gradOutput->size[3], 67 | gradOutput->size[2], 68 | gradOutput->size[0], 69 | THCudaTensor_size(state, inputImages, 1), 70 | THCudaTensor_size(state, inputImages, 2), 71 | THCudaTensor_size(state, inputImages, 3), 72 | THCudaTensor_size(state, inputImages, 0), 73 | THCudaTensor_data(state, inputImages), 74 | THCudaTensor_stride(state, inputImages, 0), 75 | THCudaTensor_stride(state, inputImages, 1), 76 | THCudaTensor_stride(state, inputImages, 2), 77 | THCudaTensor_stride(state, inputImages, 3), 78 | THCudaTensor_data(state, grids), 79 | THCudaTensor_stride(state, grids, 0), 80 | THCudaTensor_stride(state, grids, 3), 81 | THCudaTensor_stride(state, grids, 1), 82 | THCudaTensor_stride(state, grids, 2), 83 | THCudaTensor_data(state, gradInputImages), 84 | THCudaTensor_stride(state, gradInputImages, 0), 85 | THCudaTensor_stride(state, gradInputImages, 1), 86 | THCudaTensor_stride(state, gradInputImages, 2), 87 | THCudaTensor_stride(state, gradInputImages, 3), 88 | THCudaTensor_data(state, gradGrids), 89 | THCudaTensor_stride(state, gradGrids, 0), 90 | THCudaTensor_stride(state, gradGrids, 3), 91 | THCudaTensor_stride(state, gradGrids, 1), 92 | THCudaTensor_stride(state, gradGrids, 2), 93 | THCudaTensor_data(state, gradOutput), 94 | THCudaTensor_stride(state, gradOutput, 0), 95 | THCudaTensor_stride(state, gradOutput, 1), 96 | THCudaTensor_stride(state, gradOutput, 2), 97 | THCudaTensor_stride(state, gradOutput, 3), 98 | THCState_getCurrentStream(state)); 99 | 100 | //check for errors 101 | if (!success) { 102 | THError("aborting"); 103 | } 104 | return 1; 105 | } 106 | -------------------------------------------------------------------------------- /lib/model/roi_crop/src/roi_crop_cuda.h: -------------------------------------------------------------------------------- 1 | // Bilinear sampling is done in BHWD (coalescing is not obvious in BDHW) 2 | // we assume BHWD format in inputImages 3 | // we assume BHW(YX) format on grids 4 | 5 | int BilinearSamplerBHWD_updateOutput_cuda(THCudaTensor *inputImages, THCudaTensor *grids, THCudaTensor *output); 6 | 7 | int BilinearSamplerBHWD_updateGradInput_cuda(THCudaTensor *inputImages, THCudaTensor *grids, THCudaTensor *gradInputImages, 8 | THCudaTensor *gradGrids, THCudaTensor *gradOutput); 9 | -------------------------------------------------------------------------------- /lib/model/roi_crop/src/roi_crop_cuda_kernel.cu.o: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_crop/src/roi_crop_cuda_kernel.cu.o -------------------------------------------------------------------------------- /lib/model/roi_crop/src/roi_crop_cuda_kernel.h: -------------------------------------------------------------------------------- 1 | #ifdef __cplusplus 2 | extern "C" { 3 | #endif 4 | 5 | 6 | int BilinearSamplerBHWD_updateOutput_cuda_kernel(/*output->size[3]*/int oc, 7 | /*output->size[2]*/int ow, 8 | /*output->size[1]*/int oh, 9 | /*output->size[0]*/int ob, 10 | /*THCudaTensor_size(state, inputImages, 3)*/int ic, 11 | /*THCudaTensor_size(state, inputImages, 1)*/int ih, 12 | /*THCudaTensor_size(state, inputImages, 2)*/int iw, 13 | /*THCudaTensor_size(state, inputImages, 0)*/int ib, 14 | /*THCudaTensor *inputImages*/float *inputImages, int isb, int isc, int ish, int isw, 15 | /*THCudaTensor *grids*/float *grids, int gsb, int gsc, int gsh, int gsw, 16 | /*THCudaTensor *output*/float *output, int osb, int osc, int osh, int osw, 17 | /*THCState_getCurrentStream(state)*/cudaStream_t stream); 18 | 19 | int BilinearSamplerBHWD_updateGradInput_cuda_kernel(/*gradOutput->size[3]*/int goc, 20 | /*gradOutput->size[2]*/int gow, 21 | /*gradOutput->size[1]*/int goh, 22 | /*gradOutput->size[0]*/int gob, 23 | /*THCudaTensor_size(state, inputImages, 3)*/int ic, 24 | /*THCudaTensor_size(state, inputImages, 1)*/int ih, 25 | /*THCudaTensor_size(state, inputImages, 2)*/int iw, 26 | /*THCudaTensor_size(state, inputImages, 0)*/int ib, 27 | /*THCudaTensor *inputImages*/float *inputImages, int isb, int isc, int ish, int isw, 28 | /*THCudaTensor *grids*/float *grids, int gsb, int gsc, int gsh, int gsw, 29 | /*THCudaTensor *gradInputImages*/float *gradInputImages, int gisb, int gisc, int gish, int gisw, 30 | /*THCudaTensor *gradGrids*/float *gradGrids, int ggsb, int ggsc, int ggsh, int ggsw, 31 | /*THCudaTensor *gradOutput*/float *gradOutput, int gosb, int gosc, int gosh, int gosw, 32 | /*THCState_getCurrentStream(state)*/cudaStream_t stream); 33 | 34 | 35 | #ifdef __cplusplus 36 | } 37 | #endif 38 | -------------------------------------------------------------------------------- /lib/model/roi_pooling/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_pooling/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/__pycache__/__init__.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/__pycache__/__init__.cpython-36.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/_ext/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/_ext/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_pooling/_ext/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/_ext/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/_ext/__pycache__/__init__.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/_ext/__pycache__/__init__.cpython-36.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/_ext/roi_pooling/__init__.py: -------------------------------------------------------------------------------- 1 | 2 | from torch.utils.ffi import _wrap_function 3 | from ._roi_pooling import lib as _lib, ffi as _ffi 4 | 5 | __all__ = [] 6 | def _import_symbols(locals): 7 | for symbol in dir(_lib): 8 | fn = getattr(_lib, symbol) 9 | if callable(fn): 10 | locals[symbol] = _wrap_function(fn, _ffi) 11 | else: 12 | locals[symbol] = fn 13 | __all__.append(symbol) 14 | 15 | _import_symbols(locals()) 16 | -------------------------------------------------------------------------------- /lib/model/roi_pooling/_ext/roi_pooling/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/_ext/roi_pooling/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/_ext/roi_pooling/__pycache__/__init__.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/_ext/roi_pooling/__pycache__/__init__.cpython-36.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/_ext/roi_pooling/_roi_pooling.so: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/_ext/roi_pooling/_roi_pooling.so -------------------------------------------------------------------------------- /lib/model/roi_pooling/build.py: -------------------------------------------------------------------------------- 1 | from __future__ import print_function 2 | import os 3 | import torch 4 | from torch.utils.ffi import create_extension 5 | 6 | 7 | sources = ['src/roi_pooling.c'] 8 | headers = ['src/roi_pooling.h'] 9 | defines = [] 10 | with_cuda = False 11 | 12 | if torch.cuda.is_available(): 13 | print('Including CUDA code.') 14 | sources += ['src/roi_pooling_cuda.c'] 15 | headers += ['src/roi_pooling_cuda.h'] 16 | defines += [('WITH_CUDA', None)] 17 | with_cuda = True 18 | 19 | this_file = os.path.dirname(os.path.realpath(__file__)) 20 | print(this_file) 21 | extra_objects = ['src/roi_pooling.cu.o'] 22 | extra_objects = [os.path.join(this_file, fname) for fname in extra_objects] 23 | 24 | ffi = create_extension( 25 | '_ext.roi_pooling', 26 | headers=headers, 27 | sources=sources, 28 | define_macros=defines, 29 | relative_to=__file__, 30 | with_cuda=with_cuda, 31 | extra_objects=extra_objects 32 | ) 33 | 34 | if __name__ == '__main__': 35 | ffi.build() 36 | -------------------------------------------------------------------------------- /lib/model/roi_pooling/functions/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/functions/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_pooling/functions/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/functions/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/functions/__pycache__/__init__.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/functions/__pycache__/__init__.cpython-36.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/functions/__pycache__/roi_pool.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/functions/__pycache__/roi_pool.cpython-36.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/functions/roi_pool.py: -------------------------------------------------------------------------------- 1 | import torch 2 | from torch.autograd import Function 3 | from .._ext import roi_pooling 4 | import pdb 5 | 6 | class RoIPoolFunction(Function): 7 | def __init__(ctx, pooled_height, pooled_width, spatial_scale): 8 | ctx.pooled_width = pooled_width 9 | ctx.pooled_height = pooled_height 10 | ctx.spatial_scale = spatial_scale 11 | ctx.feature_size = None 12 | 13 | def forward(ctx, features, rois): 14 | ctx.feature_size = features.size() 15 | batch_size, num_channels, data_height, data_width = ctx.feature_size 16 | num_rois = rois.size(0) 17 | output = features.new(num_rois, num_channels, ctx.pooled_height, ctx.pooled_width).zero_() 18 | ctx.argmax = features.new(num_rois, num_channels, ctx.pooled_height, ctx.pooled_width).zero_().int() 19 | ctx.rois = rois 20 | if not features.is_cuda: 21 | _features = features.permute(0, 2, 3, 1) 22 | roi_pooling.roi_pooling_forward(ctx.pooled_height, ctx.pooled_width, ctx.spatial_scale, 23 | _features, rois, output) 24 | else: 25 | roi_pooling.roi_pooling_forward_cuda(ctx.pooled_height, ctx.pooled_width, ctx.spatial_scale, 26 | features, rois, output, ctx.argmax) 27 | 28 | return output 29 | 30 | def backward(ctx, grad_output): 31 | assert(ctx.feature_size is not None and grad_output.is_cuda) 32 | batch_size, num_channels, data_height, data_width = ctx.feature_size 33 | grad_input = grad_output.new(batch_size, num_channels, data_height, data_width).zero_() 34 | 35 | roi_pooling.roi_pooling_backward_cuda(ctx.pooled_height, ctx.pooled_width, ctx.spatial_scale, 36 | grad_output, ctx.rois, grad_input, ctx.argmax) 37 | 38 | return grad_input, None 39 | -------------------------------------------------------------------------------- /lib/model/roi_pooling/functions/roi_pool.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/functions/roi_pool.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/modules/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/modules/__init__.py -------------------------------------------------------------------------------- /lib/model/roi_pooling/modules/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/modules/__init__.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/modules/__pycache__/__init__.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/modules/__pycache__/__init__.cpython-36.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/modules/__pycache__/roi_pool.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/modules/__pycache__/roi_pool.cpython-36.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/modules/roi_pool.py: -------------------------------------------------------------------------------- 1 | from torch.nn.modules.module import Module 2 | from ..functions.roi_pool import RoIPoolFunction 3 | 4 | 5 | class _RoIPooling(Module): 6 | def __init__(self, pooled_height, pooled_width, spatial_scale): 7 | super(_RoIPooling, self).__init__() 8 | 9 | self.pooled_width = int(pooled_width) 10 | self.pooled_height = int(pooled_height) 11 | self.spatial_scale = float(spatial_scale) 12 | 13 | def forward(self, features, rois): 14 | return RoIPoolFunction(self.pooled_height, self.pooled_width, self.spatial_scale)(features, rois) 15 | -------------------------------------------------------------------------------- /lib/model/roi_pooling/modules/roi_pool.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/modules/roi_pool.pyc -------------------------------------------------------------------------------- /lib/model/roi_pooling/src/roi_pooling.c: -------------------------------------------------------------------------------- 1 | #include 2 | #include 3 | 4 | int roi_pooling_forward(int pooled_height, int pooled_width, float spatial_scale, 5 | THFloatTensor * features, THFloatTensor * rois, THFloatTensor * output) 6 | { 7 | // Grab the input tensor 8 | float * data_flat = THFloatTensor_data(features); 9 | float * rois_flat = THFloatTensor_data(rois); 10 | 11 | float * output_flat = THFloatTensor_data(output); 12 | 13 | // Number of ROIs 14 | int num_rois = THFloatTensor_size(rois, 0); 15 | int size_rois = THFloatTensor_size(rois, 1); 16 | // batch size 17 | int batch_size = THFloatTensor_size(features, 0); 18 | if(batch_size != 1) 19 | { 20 | return 0; 21 | } 22 | // data height 23 | int data_height = THFloatTensor_size(features, 1); 24 | // data width 25 | int data_width = THFloatTensor_size(features, 2); 26 | // Number of channels 27 | int num_channels = THFloatTensor_size(features, 3); 28 | 29 | // Set all element of the output tensor to -inf. 30 | THFloatStorage_fill(THFloatTensor_storage(output), -1); 31 | 32 | // For each ROI R = [batch_index x1 y1 x2 y2]: max pool over R 33 | int index_roi = 0; 34 | int index_output = 0; 35 | int n; 36 | for (n = 0; n < num_rois; ++n) 37 | { 38 | int roi_batch_ind = rois_flat[index_roi + 0]; 39 | int roi_start_w = round(rois_flat[index_roi + 1] * spatial_scale); 40 | int roi_start_h = round(rois_flat[index_roi + 2] * spatial_scale); 41 | int roi_end_w = round(rois_flat[index_roi + 3] * spatial_scale); 42 | int roi_end_h = round(rois_flat[index_roi + 4] * spatial_scale); 43 | // CHECK_GE(roi_batch_ind, 0); 44 | // CHECK_LT(roi_batch_ind, batch_size); 45 | 46 | int roi_height = fmaxf(roi_end_h - roi_start_h + 1, 1); 47 | int roi_width = fmaxf(roi_end_w - roi_start_w + 1, 1); 48 | float bin_size_h = (float)(roi_height) / (float)(pooled_height); 49 | float bin_size_w = (float)(roi_width) / (float)(pooled_width); 50 | 51 | int index_data = roi_batch_ind * data_height * data_width * num_channels; 52 | const int output_area = pooled_width * pooled_height; 53 | 54 | int c, ph, pw; 55 | for (ph = 0; ph < pooled_height; ++ph) 56 | { 57 | for (pw = 0; pw < pooled_width; ++pw) 58 | { 59 | int hstart = (floor((float)(ph) * bin_size_h)); 60 | int wstart = (floor((float)(pw) * bin_size_w)); 61 | int hend = (ceil((float)(ph + 1) * bin_size_h)); 62 | int wend = (ceil((float)(pw + 1) * bin_size_w)); 63 | 64 | hstart = fminf(fmaxf(hstart + roi_start_h, 0), data_height); 65 | hend = fminf(fmaxf(hend + roi_start_h, 0), data_height); 66 | wstart = fminf(fmaxf(wstart + roi_start_w, 0), data_width); 67 | wend = fminf(fmaxf(wend + roi_start_w, 0), data_width); 68 | 69 | const int pool_index = index_output + (ph * pooled_width + pw); 70 | int is_empty = (hend <= hstart) || (wend <= wstart); 71 | if (is_empty) 72 | { 73 | for (c = 0; c < num_channels * output_area; c += output_area) 74 | { 75 | output_flat[pool_index + c] = 0; 76 | } 77 | } 78 | else 79 | { 80 | int h, w, c; 81 | for (h = hstart; h < hend; ++h) 82 | { 83 | for (w = wstart; w < wend; ++w) 84 | { 85 | for (c = 0; c < num_channels; ++c) 86 | { 87 | const int index = (h * data_width + w) * num_channels + c; 88 | if (data_flat[index_data + index] > output_flat[pool_index + c * output_area]) 89 | { 90 | output_flat[pool_index + c * output_area] = data_flat[index_data + index]; 91 | } 92 | } 93 | } 94 | } 95 | } 96 | } 97 | } 98 | 99 | // Increment ROI index 100 | index_roi += size_rois; 101 | index_output += pooled_height * pooled_width * num_channels; 102 | } 103 | return 1; 104 | } -------------------------------------------------------------------------------- /lib/model/roi_pooling/src/roi_pooling.cu.o: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling/src/roi_pooling.cu.o -------------------------------------------------------------------------------- /lib/model/roi_pooling/src/roi_pooling.h: -------------------------------------------------------------------------------- 1 | int roi_pooling_forward(int pooled_height, int pooled_width, float spatial_scale, 2 | THFloatTensor * features, THFloatTensor * rois, THFloatTensor * output); -------------------------------------------------------------------------------- /lib/model/roi_pooling/src/roi_pooling_cuda.c: -------------------------------------------------------------------------------- 1 | #include 2 | #include 3 | #include "roi_pooling_kernel.h" 4 | 5 | extern THCState *state; 6 | 7 | int roi_pooling_forward_cuda(int pooled_height, int pooled_width, float spatial_scale, 8 | THCudaTensor * features, THCudaTensor * rois, THCudaTensor * output, THCudaIntTensor * argmax) 9 | { 10 | // Grab the input tensor 11 | float * data_flat = THCudaTensor_data(state, features); 12 | float * rois_flat = THCudaTensor_data(state, rois); 13 | 14 | float * output_flat = THCudaTensor_data(state, output); 15 | int * argmax_flat = THCudaIntTensor_data(state, argmax); 16 | 17 | // Number of ROIs 18 | int num_rois = THCudaTensor_size(state, rois, 0); 19 | int size_rois = THCudaTensor_size(state, rois, 1); 20 | if (size_rois != 5) 21 | { 22 | return 0; 23 | } 24 | 25 | // batch size 26 | // int batch_size = THCudaTensor_size(state, features, 0); 27 | // if (batch_size != 1) 28 | // { 29 | // return 0; 30 | // } 31 | // data height 32 | int data_height = THCudaTensor_size(state, features, 2); 33 | // data width 34 | int data_width = THCudaTensor_size(state, features, 3); 35 | // Number of channels 36 | int num_channels = THCudaTensor_size(state, features, 1); 37 | 38 | cudaStream_t stream = THCState_getCurrentStream(state); 39 | 40 | ROIPoolForwardLaucher( 41 | data_flat, spatial_scale, num_rois, data_height, 42 | data_width, num_channels, pooled_height, 43 | pooled_width, rois_flat, 44 | output_flat, argmax_flat, stream); 45 | 46 | return 1; 47 | } 48 | 49 | int roi_pooling_backward_cuda(int pooled_height, int pooled_width, float spatial_scale, 50 | THCudaTensor * top_grad, THCudaTensor * rois, THCudaTensor * bottom_grad, THCudaIntTensor * argmax) 51 | { 52 | // Grab the input tensor 53 | float * top_grad_flat = THCudaTensor_data(state, top_grad); 54 | float * rois_flat = THCudaTensor_data(state, rois); 55 | 56 | float * bottom_grad_flat = THCudaTensor_data(state, bottom_grad); 57 | int * argmax_flat = THCudaIntTensor_data(state, argmax); 58 | 59 | // Number of ROIs 60 | int num_rois = THCudaTensor_size(state, rois, 0); 61 | int size_rois = THCudaTensor_size(state, rois, 1); 62 | if (size_rois != 5) 63 | { 64 | return 0; 65 | } 66 | 67 | // batch size 68 | int batch_size = THCudaTensor_size(state, bottom_grad, 0); 69 | // if (batch_size != 1) 70 | // { 71 | // return 0; 72 | // } 73 | // data height 74 | int data_height = THCudaTensor_size(state, bottom_grad, 2); 75 | // data width 76 | int data_width = THCudaTensor_size(state, bottom_grad, 3); 77 | // Number of channels 78 | int num_channels = THCudaTensor_size(state, bottom_grad, 1); 79 | 80 | cudaStream_t stream = THCState_getCurrentStream(state); 81 | ROIPoolBackwardLaucher( 82 | top_grad_flat, spatial_scale, batch_size, num_rois, data_height, 83 | data_width, num_channels, pooled_height, 84 | pooled_width, rois_flat, 85 | bottom_grad_flat, argmax_flat, stream); 86 | 87 | return 1; 88 | } 89 | -------------------------------------------------------------------------------- /lib/model/roi_pooling/src/roi_pooling_cuda.h: -------------------------------------------------------------------------------- 1 | int roi_pooling_forward_cuda(int pooled_height, int pooled_width, float spatial_scale, 2 | THCudaTensor * features, THCudaTensor * rois, THCudaTensor * output, THCudaIntTensor * argmax); 3 | 4 | int roi_pooling_backward_cuda(int pooled_height, int pooled_width, float spatial_scale, 5 | THCudaTensor * top_grad, THCudaTensor * rois, THCudaTensor * bottom_grad, THCudaIntTensor * argmax); -------------------------------------------------------------------------------- /lib/model/roi_pooling/src/roi_pooling_kernel.h: -------------------------------------------------------------------------------- 1 | #ifndef _ROI_POOLING_KERNEL 2 | #define _ROI_POOLING_KERNEL 3 | 4 | #ifdef __cplusplus 5 | extern "C" { 6 | #endif 7 | 8 | int ROIPoolForwardLaucher( 9 | const float* bottom_data, const float spatial_scale, const int num_rois, const int height, 10 | const int width, const int channels, const int pooled_height, 11 | const int pooled_width, const float* bottom_rois, 12 | float* top_data, int* argmax_data, cudaStream_t stream); 13 | 14 | 15 | int ROIPoolBackwardLaucher(const float* top_diff, const float spatial_scale, const int batch_size, const int num_rois, 16 | const int height, const int width, const int channels, const int pooled_height, 17 | const int pooled_width, const float* bottom_rois, 18 | float* bottom_diff, const int* argmax_data, cudaStream_t stream); 19 | 20 | #ifdef __cplusplus 21 | } 22 | #endif 23 | 24 | #endif 25 | 26 | -------------------------------------------------------------------------------- /lib/model/roi_pooling_replace.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import torch.nn as nn 3 | import torch.autograd as ag 4 | import math 5 | 6 | from torch.autograd.function import Function 7 | from torch._thnn import type2backend 8 | 9 | class AdaptiveMaxPool2d(Function): 10 | def __init__(self, out_w, out_h): 11 | super(AdaptiveMaxPool2d, self).__init__() 12 | self.out_w = out_w 13 | self.out_h = out_h 14 | 15 | def forward(self, input): 16 | output = input.new() 17 | indices = input.new().long() 18 | self.save_for_backward(input) 19 | self.indices = indices 20 | self._backend = type2backend[type(input)] 21 | self._backend.SpatialAdaptiveMaxPooling_updateOutput( 22 | self._backend.library_state, input, output, indices, 23 | self.out_w, self.out_h) 24 | return output 25 | 26 | def backward(self, grad_output): 27 | input, = self.saved_tensors 28 | indices = self.indices 29 | grad_input = grad_output.new() 30 | self._backend.SpatialAdaptiveMaxPooling_updateGradInput( 31 | self._backend.library_state, input, grad_output, grad_input, 32 | indices) 33 | return grad_input, None 34 | 35 | def adaptive_max_pool(input, size): 36 | return AdaptiveMaxPool2d(size[0],size[1])(input) 37 | 38 | def roi_pooling(input, rois, size=(7,7), spatial_scale=1.0): 39 | assert(rois.dim() == 2) 40 | assert(rois.size(1) == 5) 41 | output = [] 42 | rois = rois.data.float() 43 | num_rois = rois.size(0) 44 | 45 | rois[:,1:].mul_(spatial_scale) 46 | rois = rois.long() 47 | for i in range(num_rois): 48 | roi = rois[i] 49 | im_idx = roi[0] 50 | tmp = input.narrow(0, im_idx, 1) 51 | 52 | im = tmp[..., roi[2]:(roi[4]+1), roi[1]:(roi[3]+1)] 53 | output.append(adaptive_max_pool(im, size)) 54 | 55 | return torch.cat(output, 0) 56 | 57 | if __name__ == '__main__': 58 | input = ag.Variable(torch.rand(1,1,10,10), requires_grad=True) 59 | rois = ag.Variable(torch.LongTensor([[0,1,2,7,8],[0,3,3,8,8]]),requires_grad=False) 60 | #rois = ag.Variable(torch.LongTensor([[0,3,3,8,8]]),requires_grad=False) 61 | 62 | out = adaptive_max_pool(input,(3,3)) 63 | out.backward(out.data.clone().uniform_()) 64 | 65 | out = roi_pooling(input, rois, size=(3,3)) 66 | out.backward(out.data.clone().uniform_()) 67 | 68 | -------------------------------------------------------------------------------- /lib/model/roi_pooling_replace.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/roi_pooling_replace.pyc -------------------------------------------------------------------------------- /lib/model/rpn/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/rpn/__init__.py -------------------------------------------------------------------------------- /lib/model/rpn/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/rpn/__init__.pyc -------------------------------------------------------------------------------- /lib/model/rpn/anchor_target_layer.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/rpn/anchor_target_layer.pyc -------------------------------------------------------------------------------- /lib/model/rpn/bbox_transform.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/rpn/bbox_transform.pyc -------------------------------------------------------------------------------- /lib/model/rpn/generate_anchors.py: -------------------------------------------------------------------------------- 1 | from __future__ import print_function 2 | # -------------------------------------------------------- 3 | # Faster R-CNN 4 | # Copyright (c) 2015 Microsoft 5 | # Licensed under The MIT License [see LICENSE for details] 6 | # Written by Ross Girshick and Sean Bell 7 | # -------------------------------------------------------- 8 | 9 | import numpy as np 10 | import pdb 11 | 12 | # Verify that we compute the same anchors as Shaoqing's matlab implementation: 13 | # 14 | # >> load output/rpn_cachedir/faster_rcnn_VOC2007_ZF_stage1_rpn/anchors.mat 15 | # >> anchors 16 | # 17 | # anchors = 18 | # 19 | # -83 -39 100 56 20 | # -175 -87 192 104 21 | # -359 -183 376 200 22 | # -55 -55 72 72 23 | # -119 -119 136 136 24 | # -247 -247 264 264 25 | # -35 -79 52 96 26 | # -79 -167 96 184 27 | # -167 -343 184 360 28 | 29 | #array([[ -83., -39., 100., 56.], 30 | # [-175., -87., 192., 104.], 31 | # [-359., -183., 376., 200.], 32 | # [ -55., -55., 72., 72.], 33 | # [-119., -119., 136., 136.], 34 | # [-247., -247., 264., 264.], 35 | # [ -35., -79., 52., 96.], 36 | # [ -79., -167., 96., 184.], 37 | # [-167., -343., 184., 360.]]) 38 | 39 | try: 40 | xrange # Python 2 41 | except NameError: 42 | xrange = range # Python 3 43 | 44 | 45 | def generate_anchors(base_size=16, ratios=[0.5, 1, 2], 46 | scales=2**np.arange(3, 6)): 47 | """ 48 | Generate anchor (reference) windows by enumerating aspect ratios X 49 | scales wrt a reference (0, 0, 15, 15) window. 50 | """ 51 | 52 | base_anchor = np.array([1, 1, base_size, base_size]) - 1 53 | ratio_anchors = _ratio_enum(base_anchor, ratios) 54 | anchors = np.vstack([_scale_enum(ratio_anchors[i, :], scales) 55 | for i in xrange(ratio_anchors.shape[0])]) 56 | return anchors 57 | 58 | def _whctrs(anchor): 59 | """ 60 | Return width, height, x center, and y center for an anchor (window). 61 | """ 62 | 63 | w = anchor[2] - anchor[0] + 1 64 | h = anchor[3] - anchor[1] + 1 65 | x_ctr = anchor[0] + 0.5 * (w - 1) 66 | y_ctr = anchor[1] + 0.5 * (h - 1) 67 | return w, h, x_ctr, y_ctr 68 | 69 | def _mkanchors(ws, hs, x_ctr, y_ctr): 70 | """ 71 | Given a vector of widths (ws) and heights (hs) around a center 72 | (x_ctr, y_ctr), output a set of anchors (windows). 73 | """ 74 | 75 | ws = ws[:, np.newaxis] 76 | hs = hs[:, np.newaxis] 77 | anchors = np.hstack((x_ctr - 0.5 * (ws - 1), 78 | y_ctr - 0.5 * (hs - 1), 79 | x_ctr + 0.5 * (ws - 1), 80 | y_ctr + 0.5 * (hs - 1))) 81 | return anchors 82 | 83 | def _ratio_enum(anchor, ratios): 84 | """ 85 | Enumerate a set of anchors for each aspect ratio wrt an anchor. 86 | """ 87 | 88 | w, h, x_ctr, y_ctr = _whctrs(anchor) 89 | size = w * h 90 | size_ratios = size / ratios 91 | ws = np.round(np.sqrt(size_ratios)) 92 | hs = np.round(ws * ratios) 93 | anchors = _mkanchors(ws, hs, x_ctr, y_ctr) 94 | return anchors 95 | 96 | def _scale_enum(anchor, scales): 97 | """ 98 | Enumerate a set of anchors for each scale wrt an anchor. 99 | """ 100 | 101 | w, h, x_ctr, y_ctr = _whctrs(anchor) 102 | ws = w * scales 103 | hs = h * scales 104 | anchors = _mkanchors(ws, hs, x_ctr, y_ctr) 105 | return anchors 106 | 107 | if __name__ == '__main__': 108 | import time 109 | t = time.time() 110 | a = generate_anchors() 111 | print(time.time() - t) 112 | print(a) 113 | from IPython import embed; embed() 114 | -------------------------------------------------------------------------------- /lib/model/rpn/generate_anchors.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/rpn/generate_anchors.pyc -------------------------------------------------------------------------------- /lib/model/rpn/proposal_layer.py: -------------------------------------------------------------------------------- 1 | from __future__ import absolute_import 2 | # -------------------------------------------------------- 3 | # Faster R-CNN 4 | # Copyright (c) 2015 Microsoft 5 | # Licensed under The MIT License [see LICENSE for details] 6 | # Written by Ross Girshick and Sean Bell 7 | # -------------------------------------------------------- 8 | # -------------------------------------------------------- 9 | # Reorganized and modified by Jianwei Yang and Jiasen Lu 10 | # -------------------------------------------------------- 11 | 12 | import torch 13 | import torch.nn as nn 14 | import numpy as np 15 | import math 16 | import yaml 17 | from model.utils.config import cfg 18 | from .generate_anchors import generate_anchors 19 | from .bbox_transform import bbox_transform_inv, clip_boxes, clip_boxes_batch 20 | from model.nms.nms_wrapper import nms 21 | 22 | import pdb 23 | 24 | DEBUG = False 25 | 26 | class _ProposalLayer(nn.Module): 27 | """ 28 | Outputs object detection proposals by applying estimated bounding-box 29 | transformations to a set of regular boxes (called "anchors"). 30 | """ 31 | 32 | def __init__(self, feat_stride, scales, ratios): 33 | super(_ProposalLayer, self).__init__() 34 | 35 | self._feat_stride = feat_stride 36 | self._anchors = torch.from_numpy(generate_anchors(scales=np.array(scales), 37 | ratios=np.array(ratios))).float() 38 | self._num_anchors = self._anchors.size(0) 39 | 40 | # rois blob: holds R regions of interest, each is a 5-tuple 41 | # (n, x1, y1, x2, y2) specifying an image batch index n and a 42 | # rectangle (x1, y1, x2, y2) 43 | # top[0].reshape(1, 5) 44 | # 45 | # # scores blob: holds scores for R regions of interest 46 | # if len(top) > 1: 47 | # top[1].reshape(1, 1, 1, 1) 48 | 49 | def forward(self, input): 50 | 51 | # Algorithm: 52 | # 53 | # for each (H, W) location i 54 | # generate A anchor boxes centered on cell i 55 | # apply predicted bbox deltas at cell i to each of the A anchors 56 | # clip predicted boxes to image 57 | # remove predicted boxes with either height or width < threshold 58 | # sort all (proposal, score) pairs by score from highest to lowest 59 | # take top pre_nms_topN proposals before NMS 60 | # apply NMS with threshold 0.7 to remaining proposals 61 | # take after_nms_topN proposals after NMS 62 | # return the top proposals (-> RoIs top, scores top) 63 | 64 | 65 | # the first set of _num_anchors channels are bg probs 66 | # the second set are the fg probs 67 | scores = input[0][:, self._num_anchors:, :, :] 68 | bbox_deltas = input[1] 69 | im_info = input[2] 70 | cfg_key = input[3] 71 | 72 | pre_nms_topN = cfg[cfg_key].RPN_PRE_NMS_TOP_N 73 | post_nms_topN = cfg[cfg_key].RPN_POST_NMS_TOP_N 74 | nms_thresh = cfg[cfg_key].RPN_NMS_THRESH 75 | min_size = cfg[cfg_key].RPN_MIN_SIZE 76 | 77 | batch_size = bbox_deltas.size(0) 78 | 79 | feat_height, feat_width = scores.size(2), scores.size(3) 80 | shift_x = np.arange(0, feat_width) * self._feat_stride 81 | shift_y = np.arange(0, feat_height) * self._feat_stride 82 | shift_x, shift_y = np.meshgrid(shift_x, shift_y) 83 | shifts = torch.from_numpy(np.vstack((shift_x.ravel(), shift_y.ravel(), 84 | shift_x.ravel(), shift_y.ravel())).transpose()) 85 | shifts = shifts.contiguous().type_as(scores).float() 86 | 87 | A = self._num_anchors 88 | K = shifts.size(0) 89 | 90 | self._anchors = self._anchors.type_as(scores) 91 | # anchors = self._anchors.view(1, A, 4) + shifts.view(1, K, 4).permute(1, 0, 2).contiguous() 92 | anchors = self._anchors.view(1, A, 4) + shifts.view(K, 1, 4) 93 | anchors = anchors.view(1, K * A, 4).expand(batch_size, K * A, 4) 94 | 95 | # Transpose and reshape predicted bbox transformations to get them 96 | # into the same order as the anchors: 97 | 98 | bbox_deltas = bbox_deltas.permute(0, 2, 3, 1).contiguous() 99 | bbox_deltas = bbox_deltas.view(batch_size, -1, 4) 100 | 101 | # Same story for the scores: 102 | scores = scores.permute(0, 2, 3, 1).contiguous() 103 | scores = scores.view(batch_size, -1) 104 | 105 | # Convert anchors into proposals via bbox transformations 106 | proposals = bbox_transform_inv(anchors, bbox_deltas, batch_size) 107 | 108 | # 2. clip predicted boxes to image 109 | proposals = clip_boxes(proposals, im_info, batch_size) 110 | # proposals = clip_boxes_batch(proposals, im_info, batch_size) 111 | 112 | # assign the score to 0 if it's non keep. 113 | # keep = self._filter_boxes(proposals, min_size * im_info[:, 2]) 114 | 115 | # trim keep index to make it euqal over batch 116 | # keep_idx = torch.cat(tuple(keep_idx), 0) 117 | 118 | # scores_keep = scores.view(-1)[keep_idx].view(batch_size, trim_size) 119 | # proposals_keep = proposals.view(-1, 4)[keep_idx, :].contiguous().view(batch_size, trim_size, 4) 120 | 121 | # _, order = torch.sort(scores_keep, 1, True) 122 | 123 | scores_keep = scores 124 | proposals_keep = proposals 125 | _, order = torch.sort(scores_keep, 1, True) 126 | 127 | output = scores.new(batch_size, post_nms_topN, 5).zero_() 128 | for i in range(batch_size): 129 | # # 3. remove predicted boxes with either height or width < threshold 130 | # # (NOTE: convert min_size to input image scale stored in im_info[2]) 131 | proposals_single = proposals_keep[i] 132 | scores_single = scores_keep[i] 133 | 134 | # # 4. sort all (proposal, score) pairs by score from highest to lowest 135 | # # 5. take top pre_nms_topN (e.g. 6000) 136 | order_single = order[i] 137 | 138 | if pre_nms_topN > 0 and pre_nms_topN < scores_keep.numel(): 139 | order_single = order_single[:pre_nms_topN] 140 | 141 | proposals_single = proposals_single[order_single, :] 142 | scores_single = scores_single[order_single].view(-1,1) 143 | 144 | # 6. apply nms (e.g. threshold = 0.7) 145 | # 7. take after_nms_topN (e.g. 300) 146 | # 8. return the top proposals (-> RoIs top) 147 | 148 | keep_idx_i = nms(torch.cat((proposals_single, scores_single), 1), nms_thresh) 149 | keep_idx_i = keep_idx_i.long().view(-1) 150 | 151 | if post_nms_topN > 0: 152 | keep_idx_i = keep_idx_i[:post_nms_topN] 153 | proposals_single = proposals_single[keep_idx_i, :] 154 | scores_single = scores_single[keep_idx_i, :] 155 | 156 | # padding 0 at the end. 157 | num_proposal = proposals_single.size(0) 158 | output[i,:,0] = i 159 | output[i,:num_proposal,1:] = proposals_single 160 | 161 | return output 162 | 163 | def backward(self, top, propagate_down, bottom): 164 | """This layer does not propagate gradients.""" 165 | pass 166 | 167 | def reshape(self, bottom, top): 168 | """Reshaping happens during the call to forward.""" 169 | pass 170 | 171 | def _filter_boxes(self, boxes, min_size): 172 | """Remove all boxes with any side smaller than min_size.""" 173 | ws = boxes[:, :, 2] - boxes[:, :, 0] + 1 174 | hs = boxes[:, :, 3] - boxes[:, :, 1] + 1 175 | keep = ((ws >= min_size.view(-1,1).expand_as(ws)) & (hs >= min_size.view(-1,1).expand_as(hs))) 176 | return keep 177 | -------------------------------------------------------------------------------- /lib/model/rpn/proposal_layer.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/rpn/proposal_layer.pyc -------------------------------------------------------------------------------- /lib/model/rpn/proposal_target_layer_cascade.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/rpn/proposal_target_layer_cascade.pyc -------------------------------------------------------------------------------- /lib/model/rpn/rpn.py: -------------------------------------------------------------------------------- 1 | from __future__ import absolute_import 2 | import torch 3 | import torch.nn as nn 4 | import torch.nn.functional as F 5 | from torch.autograd import Variable 6 | 7 | from model.utils.config import cfg 8 | from .proposal_layer import _ProposalLayer 9 | from .anchor_target_layer import _AnchorTargetLayer 10 | from model.utils.net_utils import _smooth_l1_loss 11 | 12 | import numpy as np 13 | import math 14 | import pdb 15 | import time 16 | 17 | class _RPN(nn.Module): 18 | """ region proposal network """ 19 | def __init__(self, din): 20 | super(_RPN, self).__init__() 21 | 22 | self.din = din # get depth of input feature map, e.g., 512 23 | self.anchor_scales = cfg.ANCHOR_SCALES 24 | self.anchor_ratios = cfg.ANCHOR_RATIOS 25 | self.feat_stride = cfg.FEAT_STRIDE[0] 26 | 27 | # define the convrelu layers processing input feature map 28 | self.RPN_Conv = nn.Conv2d(self.din, 512, 3, 1, 1, bias=True) 29 | 30 | # define bg/fg classifcation score layer 31 | self.nc_score_out = len(self.anchor_scales) * len(self.anchor_ratios) * 2 # 2(bg/fg) * 9 (anchors) 32 | self.RPN_cls_score = nn.Conv2d(512, self.nc_score_out, 1, 1, 0) 33 | 34 | # define anchor box offset prediction layer 35 | self.nc_bbox_out = len(self.anchor_scales) * len(self.anchor_ratios) * 4 # 4(coords) * 9 (anchors) 36 | self.RPN_bbox_pred = nn.Conv2d(512, self.nc_bbox_out, 1, 1, 0) 37 | 38 | # define proposal layer 39 | self.RPN_proposal = _ProposalLayer(self.feat_stride, self.anchor_scales, self.anchor_ratios) 40 | 41 | # define anchor target layer 42 | self.RPN_anchor_target = _AnchorTargetLayer(self.feat_stride, self.anchor_scales, self.anchor_ratios) 43 | 44 | self.rpn_loss_cls = 0 45 | self.rpn_loss_box = 0 46 | 47 | @staticmethod 48 | def reshape(x, d): 49 | input_shape = x.size() 50 | x = x.view( 51 | input_shape[0], 52 | int(d), 53 | int(float(input_shape[1] * input_shape[2]) / float(d)), 54 | input_shape[3] 55 | ) 56 | return x 57 | 58 | def forward(self, base_feat, im_info, gt_boxes, num_boxes): 59 | 60 | batch_size = base_feat.size(0) 61 | 62 | # return feature map after convrelu layer 63 | rpn_conv1 = F.relu(self.RPN_Conv(base_feat), inplace=True) 64 | # get rpn classification score 65 | rpn_cls_score = self.RPN_cls_score(rpn_conv1) 66 | 67 | rpn_cls_score_reshape = self.reshape(rpn_cls_score, 2) 68 | rpn_cls_prob_reshape = F.softmax(rpn_cls_score_reshape) 69 | rpn_cls_prob = self.reshape(rpn_cls_prob_reshape, self.nc_score_out) 70 | 71 | # get rpn offsets to the anchor boxes 72 | rpn_bbox_pred = self.RPN_bbox_pred(rpn_conv1) 73 | 74 | # proposal layer 75 | cfg_key = 'TRAIN' if self.training else 'TEST' 76 | 77 | rois = self.RPN_proposal((rpn_cls_prob.data, rpn_bbox_pred.data, 78 | im_info, cfg_key)) 79 | 80 | self.rpn_loss_cls = 0 81 | self.rpn_loss_box = 0 82 | 83 | # generating training labels and build the rpn loss 84 | if self.training: 85 | assert gt_boxes is not None 86 | 87 | rpn_data = self.RPN_anchor_target((rpn_cls_score.data, gt_boxes, im_info, num_boxes)) 88 | 89 | # compute classification loss 90 | rpn_cls_score = rpn_cls_score_reshape.permute(0, 2, 3, 1).contiguous().view(batch_size, -1, 2) 91 | rpn_label = rpn_data[0].view(batch_size, -1) 92 | 93 | rpn_keep = Variable(rpn_label.view(-1).ne(-1).nonzero().view(-1)) 94 | rpn_cls_score = torch.index_select(rpn_cls_score.view(-1,2), 0, rpn_keep) 95 | rpn_label = torch.index_select(rpn_label.view(-1), 0, rpn_keep.data) 96 | rpn_label = Variable(rpn_label.long()) 97 | self.rpn_loss_cls = F.cross_entropy(rpn_cls_score, rpn_label) 98 | fg_cnt = torch.sum(rpn_label.data.ne(0)) 99 | 100 | rpn_bbox_targets, rpn_bbox_inside_weights, rpn_bbox_outside_weights = rpn_data[1:] 101 | 102 | # compute bbox regression loss 103 | rpn_bbox_inside_weights = Variable(rpn_bbox_inside_weights) 104 | rpn_bbox_outside_weights = Variable(rpn_bbox_outside_weights) 105 | rpn_bbox_targets = Variable(rpn_bbox_targets) 106 | 107 | self.rpn_loss_box = _smooth_l1_loss(rpn_bbox_pred, rpn_bbox_targets, rpn_bbox_inside_weights, 108 | rpn_bbox_outside_weights, sigma=3, dim=[1,2,3]) 109 | 110 | return rois, self.rpn_loss_cls, self.rpn_loss_box 111 | -------------------------------------------------------------------------------- /lib/model/rpn/rpn.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/rpn/rpn.pyc -------------------------------------------------------------------------------- /lib/model/utils/.gitignore: -------------------------------------------------------------------------------- 1 | *.c 2 | *.cpp 3 | *.so 4 | -------------------------------------------------------------------------------- /lib/model/utils/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/utils/__init__.py -------------------------------------------------------------------------------- /lib/model/utils/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/utils/__init__.pyc -------------------------------------------------------------------------------- /lib/model/utils/__pycache__/__init__.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/utils/__pycache__/__init__.cpython-36.pyc -------------------------------------------------------------------------------- /lib/model/utils/__pycache__/config.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/utils/__pycache__/config.cpython-36.pyc -------------------------------------------------------------------------------- /lib/model/utils/bbox.pyx: -------------------------------------------------------------------------------- 1 | # -------------------------------------------------------- 2 | # Fast R-CNN 3 | # Copyright (c) 2015 Microsoft 4 | # Licensed under The MIT License [see LICENSE for details] 5 | # Written by Sergey Karayev 6 | # -------------------------------------------------------- 7 | 8 | cimport cython 9 | import numpy as np 10 | cimport numpy as np 11 | 12 | DTYPE = np.float 13 | ctypedef np.float_t DTYPE_t 14 | 15 | def bbox_overlaps(np.ndarray[DTYPE_t, ndim=2] boxes, 16 | np.ndarray[DTYPE_t, ndim=2] query_boxes): 17 | return bbox_overlaps_c(boxes, query_boxes) 18 | 19 | cdef np.ndarray[DTYPE_t, ndim=2] bbox_overlaps_c( 20 | np.ndarray[DTYPE_t, ndim=2] boxes, 21 | np.ndarray[DTYPE_t, ndim=2] query_boxes): 22 | """ 23 | Parameters 24 | ---------- 25 | boxes: (N, 4) ndarray of float 26 | query_boxes: (K, 4) ndarray of float 27 | Returns 28 | ------- 29 | overlaps: (N, K) ndarray of overlap between boxes and query_boxes 30 | """ 31 | cdef unsigned int N = boxes.shape[0] 32 | cdef unsigned int K = query_boxes.shape[0] 33 | cdef np.ndarray[DTYPE_t, ndim=2] overlaps = np.zeros((N, K), dtype=DTYPE) 34 | cdef DTYPE_t iw, ih, box_area 35 | cdef DTYPE_t ua 36 | cdef unsigned int k, n 37 | for k in range(K): 38 | box_area = ( 39 | (query_boxes[k, 2] - query_boxes[k, 0] + 1) * 40 | (query_boxes[k, 3] - query_boxes[k, 1] + 1) 41 | ) 42 | for n in range(N): 43 | iw = ( 44 | min(boxes[n, 2], query_boxes[k, 2]) - 45 | max(boxes[n, 0], query_boxes[k, 0]) + 1 46 | ) 47 | if iw > 0: 48 | ih = ( 49 | min(boxes[n, 3], query_boxes[k, 3]) - 50 | max(boxes[n, 1], query_boxes[k, 1]) + 1 51 | ) 52 | if ih > 0: 53 | ua = float( 54 | (boxes[n, 2] - boxes[n, 0] + 1) * 55 | (boxes[n, 3] - boxes[n, 1] + 1) + 56 | box_area - iw * ih 57 | ) 58 | overlaps[n, k] = iw * ih / ua 59 | return overlaps 60 | 61 | 62 | def bbox_intersections( 63 | np.ndarray[DTYPE_t, ndim=2] boxes, 64 | np.ndarray[DTYPE_t, ndim=2] query_boxes): 65 | return bbox_intersections_c(boxes, query_boxes) 66 | 67 | 68 | cdef np.ndarray[DTYPE_t, ndim=2] bbox_intersections_c( 69 | np.ndarray[DTYPE_t, ndim=2] boxes, 70 | np.ndarray[DTYPE_t, ndim=2] query_boxes): 71 | """ 72 | For each query box compute the intersection ratio covered by boxes 73 | ---------- 74 | Parameters 75 | ---------- 76 | boxes: (N, 4) ndarray of float 77 | query_boxes: (K, 4) ndarray of float 78 | Returns 79 | ------- 80 | overlaps: (N, K) ndarray of intersec between boxes and query_boxes 81 | """ 82 | cdef unsigned int N = boxes.shape[0] 83 | cdef unsigned int K = query_boxes.shape[0] 84 | cdef np.ndarray[DTYPE_t, ndim=2] intersec = np.zeros((N, K), dtype=DTYPE) 85 | cdef DTYPE_t iw, ih, box_area 86 | cdef DTYPE_t ua 87 | cdef unsigned int k, n 88 | for k in range(K): 89 | box_area = ( 90 | (query_boxes[k, 2] - query_boxes[k, 0] + 1) * 91 | (query_boxes[k, 3] - query_boxes[k, 1] + 1) 92 | ) 93 | for n in range(N): 94 | iw = ( 95 | min(boxes[n, 2], query_boxes[k, 2]) - 96 | max(boxes[n, 0], query_boxes[k, 0]) + 1 97 | ) 98 | if iw > 0: 99 | ih = ( 100 | min(boxes[n, 3], query_boxes[k, 3]) - 101 | max(boxes[n, 1], query_boxes[k, 1]) + 1 102 | ) 103 | if ih > 0: 104 | intersec[n, k] = iw * ih / box_area 105 | return intersec -------------------------------------------------------------------------------- /lib/model/utils/blob.py: -------------------------------------------------------------------------------- 1 | # -------------------------------------------------------- 2 | # Fast R-CNN 3 | # Copyright (c) 2015 Microsoft 4 | # Licensed under The MIT License [see LICENSE for details] 5 | # Written by Ross Girshick 6 | # -------------------------------------------------------- 7 | 8 | """Blob helper functions.""" 9 | 10 | import numpy as np 11 | # from scipy.misc import imread, imresize 12 | import cv2 13 | 14 | try: 15 | xrange # Python 2 16 | except NameError: 17 | xrange = range # Python 3 18 | 19 | 20 | def im_list_to_blob(ims): 21 | """Convert a list of images into a network input. 22 | 23 | Assumes images are already prepared (means subtracted, BGR order, ...). 24 | """ 25 | max_shape = np.array([im.shape for im in ims]).max(axis=0) 26 | num_images = len(ims) 27 | blob = np.zeros((num_images, max_shape[0], max_shape[1], 3), 28 | dtype=np.float32) 29 | for i in xrange(num_images): 30 | im = ims[i] 31 | blob[i, 0:im.shape[0], 0:im.shape[1], :] = im 32 | 33 | return blob 34 | 35 | def prep_im_for_blob(im, pixel_means, target_size, max_size): 36 | """Mean subtract and scale an image for use in a blob.""" 37 | 38 | im = im.astype(np.float32, copy=False) 39 | im -= pixel_means 40 | # im = im[:, :, ::-1] 41 | im_shape = im.shape 42 | im_size_min = np.min(im_shape[0:2]) 43 | im_size_max = np.max(im_shape[0:2]) 44 | im_scale = float(target_size) / float(im_size_min) 45 | # Prevent the biggest axis from being more than MAX_SIZE 46 | # if np.round(im_scale * im_size_max) > max_size: 47 | # im_scale = float(max_size) / float(im_size_max) 48 | # im = imresize(im, im_scale) 49 | im = cv2.resize(im, None, None, fx=im_scale, fy=im_scale, 50 | interpolation=cv2.INTER_LINEAR) 51 | 52 | return im, im_scale 53 | -------------------------------------------------------------------------------- /lib/model/utils/blob.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/utils/blob.pyc -------------------------------------------------------------------------------- /lib/model/utils/config.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/utils/config.pyc -------------------------------------------------------------------------------- /lib/model/utils/cython_bbox.cpython-36m-x86_64-linux-gnu.so: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/utils/cython_bbox.cpython-36m-x86_64-linux-gnu.so -------------------------------------------------------------------------------- /lib/model/utils/cython_bbox.so: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/utils/cython_bbox.so -------------------------------------------------------------------------------- /lib/model/utils/logger.py: -------------------------------------------------------------------------------- 1 | # Code referenced from https://gist.github.com/gyglim/1f8dfb1b5c82627ae3efcfbbadb9f514 2 | import tensorflow as tf 3 | import numpy as np 4 | import scipy.misc 5 | try: 6 | from StringIO import StringIO # Python 2.7 7 | except ImportError: 8 | from io import BytesIO # Python 3.x 9 | 10 | 11 | class Logger(object): 12 | 13 | def __init__(self, log_dir): 14 | """Create a summary writer logging to log_dir.""" 15 | self.writer = tf.summary.FileWriter(log_dir) 16 | 17 | def scalar_summary(self, tag, value, step): 18 | """Log a scalar variable.""" 19 | summary = tf.Summary(value=[tf.Summary.Value(tag=tag, simple_value=value)]) 20 | self.writer.add_summary(summary, step) 21 | 22 | def image_summary(self, tag, images, step): 23 | """Log a list of images.""" 24 | 25 | img_summaries = [] 26 | for i, img in enumerate(images): 27 | # Write the image to a string 28 | try: 29 | s = StringIO() 30 | except: 31 | s = BytesIO() 32 | scipy.misc.toimage(img).save(s, format="png") 33 | 34 | # Create an Image object 35 | img_sum = tf.Summary.Image(encoded_image_string=s.getvalue(), 36 | height=img.shape[0], 37 | width=img.shape[1]) 38 | # Create a Summary value 39 | img_summaries.append(tf.Summary.Value(tag='%s/%d' % (tag, i), image=img_sum)) 40 | 41 | # Create and write Summary 42 | summary = tf.Summary(value=img_summaries) 43 | self.writer.add_summary(summary, step) 44 | 45 | def histo_summary(self, tag, values, step, bins=1000): 46 | """Log a histogram of the tensor of values.""" 47 | 48 | # Create a histogram using numpy 49 | counts, bin_edges = np.histogram(values, bins=bins) 50 | 51 | # Fill the fields of the histogram proto 52 | hist = tf.HistogramProto() 53 | hist.min = float(np.min(values)) 54 | hist.max = float(np.max(values)) 55 | hist.num = int(np.prod(values.shape)) 56 | hist.sum = float(np.sum(values)) 57 | hist.sum_squares = float(np.sum(values**2)) 58 | 59 | # Drop the start of the first bin 60 | bin_edges = bin_edges[1:] 61 | 62 | # Add bin edges and counts 63 | for edge in bin_edges: 64 | hist.bucket_limit.append(edge) 65 | for c in counts: 66 | hist.bucket.append(c) 67 | 68 | # Create and write Summary 69 | summary = tf.Summary(value=[tf.Summary.Value(tag=tag, histo=hist)]) 70 | self.writer.add_summary(summary, step) 71 | self.writer.flush() 72 | -------------------------------------------------------------------------------- /lib/model/utils/net_utils.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/model/utils/net_utils.pyc -------------------------------------------------------------------------------- /lib/pycocotools/UPSTREAM_REV: -------------------------------------------------------------------------------- 1 | https://github.com/pdollar/coco/commit/3ac47c77ebd5a1ed4254a98b7fbf2ef4765a3574 2 | -------------------------------------------------------------------------------- /lib/pycocotools/__init__.py: -------------------------------------------------------------------------------- 1 | __author__ = 'tylin' 2 | -------------------------------------------------------------------------------- /lib/pycocotools/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/pycocotools/__init__.pyc -------------------------------------------------------------------------------- /lib/pycocotools/_mask.cpython-36m-x86_64-linux-gnu.so: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/pycocotools/_mask.cpython-36m-x86_64-linux-gnu.so -------------------------------------------------------------------------------- /lib/pycocotools/_mask.so: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/pycocotools/_mask.so -------------------------------------------------------------------------------- /lib/pycocotools/coco.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/pycocotools/coco.pyc -------------------------------------------------------------------------------- /lib/pycocotools/cocoeval.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/pycocotools/cocoeval.pyc -------------------------------------------------------------------------------- /lib/pycocotools/license.txt: -------------------------------------------------------------------------------- 1 | Copyright (c) 2014, Piotr Dollar and Tsung-Yi Lin 2 | All rights reserved. 3 | 4 | Redistribution and use in source and binary forms, with or without 5 | modification, are permitted provided that the following conditions are met: 6 | 7 | 1. Redistributions of source code must retain the above copyright notice, this 8 | list of conditions and the following disclaimer. 9 | 2. Redistributions in binary form must reproduce the above copyright notice, 10 | this list of conditions and the following disclaimer in the documentation 11 | and/or other materials provided with the distribution. 12 | 13 | THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND 14 | ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 15 | WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 16 | DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR 17 | ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 18 | (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 19 | LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND 20 | ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 21 | (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 22 | SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 23 | 24 | The views and conclusions contained in the software and documentation are those 25 | of the authors and should not be interpreted as representing official policies, 26 | either expressed or implied, of the FreeBSD Project. 27 | -------------------------------------------------------------------------------- /lib/pycocotools/mask.py: -------------------------------------------------------------------------------- 1 | __author__ = 'tsungyi' 2 | 3 | from . import _mask 4 | 5 | # Interface for manipulating masks stored in RLE format. 6 | # 7 | # RLE is a simple yet efficient format for storing binary masks. RLE 8 | # first divides a vector (or vectorized image) into a series of piecewise 9 | # constant regions and then for each piece simply stores the length of 10 | # that piece. For example, given M=[0 0 1 1 1 0 1] the RLE counts would 11 | # be [2 3 1 1], or for M=[1 1 1 1 1 1 0] the counts would be [0 6 1] 12 | # (note that the odd counts are always the numbers of zeros). Instead of 13 | # storing the counts directly, additional compression is achieved with a 14 | # variable bitrate representation based on a common scheme called LEB128. 15 | # 16 | # Compression is greatest given large piecewise constant regions. 17 | # Specifically, the size of the RLE is proportional to the number of 18 | # *boundaries* in M (or for an image the number of boundaries in the y 19 | # direction). Assuming fairly simple shapes, the RLE representation is 20 | # O(sqrt(n)) where n is number of pixels in the object. Hence space usage 21 | # is substantially lower, especially for large simple objects (large n). 22 | # 23 | # Many common operations on masks can be computed directly using the RLE 24 | # (without need for decoding). This includes computations such as area, 25 | # union, intersection, etc. All of these operations are linear in the 26 | # size of the RLE, in other words they are O(sqrt(n)) where n is the area 27 | # of the object. Computing these operations on the original mask is O(n). 28 | # Thus, using the RLE can result in substantial computational savings. 29 | # 30 | # The following API functions are defined: 31 | # encode - Encode binary masks using RLE. 32 | # decode - Decode binary masks encoded via RLE. 33 | # merge - Compute union or intersection of encoded masks. 34 | # iou - Compute intersection over union between masks. 35 | # area - Compute area of encoded masks. 36 | # toBbox - Get bounding boxes surrounding encoded masks. 37 | # frPyObjects - Convert polygon, bbox, and uncompressed RLE to encoded RLE mask. 38 | # 39 | # Usage: 40 | # Rs = encode( masks ) 41 | # masks = decode( Rs ) 42 | # R = merge( Rs, intersect=false ) 43 | # o = iou( dt, gt, iscrowd ) 44 | # a = area( Rs ) 45 | # bbs = toBbox( Rs ) 46 | # Rs = frPyObjects( [pyObjects], h, w ) 47 | # 48 | # In the API the following formats are used: 49 | # Rs - [dict] Run-length encoding of binary masks 50 | # R - dict Run-length encoding of binary mask 51 | # masks - [hxwxn] Binary mask(s) (must have type np.ndarray(dtype=uint8) in column-major order) 52 | # iscrowd - [nx1] list of np.ndarray. 1 indicates corresponding gt image has crowd region to ignore 53 | # bbs - [nx4] Bounding box(es) stored as [x y w h] 54 | # poly - Polygon stored as [[x1 y1 x2 y2...],[x1 y1 ...],...] (2D list) 55 | # dt,gt - May be either bounding boxes or encoded masks 56 | # Both poly and bbs are 0-indexed (bbox=[0 0 1 1] encloses first pixel). 57 | # 58 | # Finally, a note about the intersection over union (iou) computation. 59 | # The standard iou of a ground truth (gt) and detected (dt) object is 60 | # iou(gt,dt) = area(intersect(gt,dt)) / area(union(gt,dt)) 61 | # For "crowd" regions, we use a modified criteria. If a gt object is 62 | # marked as "iscrowd", we allow a dt to match any subregion of the gt. 63 | # Choosing gt' in the crowd gt that best matches the dt can be done using 64 | # gt'=intersect(dt,gt). Since by definition union(gt',dt)=dt, computing 65 | # iou(gt,dt,iscrowd) = iou(gt',dt) = area(intersect(gt,dt)) / area(dt) 66 | # For crowd gt regions we use this modified criteria above for the iou. 67 | # 68 | # To compile run "python setup.py build_ext --inplace" 69 | # Please do not contact us for help with compiling. 70 | # 71 | # Microsoft COCO Toolbox. version 2.0 72 | # Data, paper, and tutorials available at: http://mscoco.org/ 73 | # Code written by Piotr Dollar and Tsung-Yi Lin, 2015. 74 | # Licensed under the Simplified BSD License [see coco/license.txt] 75 | 76 | encode = _mask.encode 77 | decode = _mask.decode 78 | iou = _mask.iou 79 | merge = _mask.merge 80 | area = _mask.area 81 | toBbox = _mask.toBbox 82 | frPyObjects = _mask.frPyObjects -------------------------------------------------------------------------------- /lib/pycocotools/mask.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/pycocotools/mask.pyc -------------------------------------------------------------------------------- /lib/pycocotools/maskApi.h: -------------------------------------------------------------------------------- 1 | /************************************************************************** 2 | * Microsoft COCO Toolbox. version 2.0 3 | * Data, paper, and tutorials available at: http://mscoco.org/ 4 | * Code written by Piotr Dollar and Tsung-Yi Lin, 2015. 5 | * Licensed under the Simplified BSD License [see coco/license.txt] 6 | **************************************************************************/ 7 | #pragma once 8 | #include 9 | 10 | typedef unsigned int uint; 11 | typedef unsigned long siz; 12 | typedef unsigned char byte; 13 | typedef double* BB; 14 | typedef struct { siz h, w, m; uint *cnts; } RLE; 15 | 16 | // Initialize/destroy RLE. 17 | void rleInit( RLE *R, siz h, siz w, siz m, uint *cnts ); 18 | void rleFree( RLE *R ); 19 | 20 | // Initialize/destroy RLE array. 21 | void rlesInit( RLE **R, siz n ); 22 | void rlesFree( RLE **R, siz n ); 23 | 24 | // Encode binary masks using RLE. 25 | void rleEncode( RLE *R, const byte *mask, siz h, siz w, siz n ); 26 | 27 | // Decode binary masks encoded via RLE. 28 | void rleDecode( const RLE *R, byte *mask, siz n ); 29 | 30 | // Compute union or intersection of encoded masks. 31 | void rleMerge( const RLE *R, RLE *M, siz n, bool intersect ); 32 | 33 | // Compute area of encoded masks. 34 | void rleArea( const RLE *R, siz n, uint *a ); 35 | 36 | // Compute intersection over union between masks. 37 | void rleIou( RLE *dt, RLE *gt, siz m, siz n, byte *iscrowd, double *o ); 38 | 39 | // Compute intersection over union between bounding boxes. 40 | void bbIou( BB dt, BB gt, siz m, siz n, byte *iscrowd, double *o ); 41 | 42 | // Get bounding boxes surrounding encoded masks. 43 | void rleToBbox( const RLE *R, BB bb, siz n ); 44 | 45 | // Convert bounding boxes to encoded masks. 46 | void rleFrBbox( RLE *R, const BB bb, siz h, siz w, siz n ); 47 | 48 | // Convert polygon to encoded mask. 49 | void rleFrPoly( RLE *R, const double *xy, siz k, siz h, siz w ); 50 | 51 | // Get compressed string representation of encoded mask. 52 | char* rleToString( const RLE *R ); 53 | 54 | // Convert from compressed string representation of encoded mask. 55 | void rleFrString( RLE *R, char *s, siz h, siz w ); 56 | -------------------------------------------------------------------------------- /lib/roi_data_layer/__init__.py: -------------------------------------------------------------------------------- 1 | # -------------------------------------------------------- 2 | # Fast R-CNN 3 | # Copyright (c) 2015 Microsoft 4 | # Licensed under The MIT License [see LICENSE for details] 5 | # Written by Ross Girshick 6 | # -------------------------------------------------------- 7 | -------------------------------------------------------------------------------- /lib/roi_data_layer/__init__.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/roi_data_layer/__init__.pyc -------------------------------------------------------------------------------- /lib/roi_data_layer/__pycache__/__init__.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/roi_data_layer/__pycache__/__init__.cpython-36.pyc -------------------------------------------------------------------------------- /lib/roi_data_layer/__pycache__/roidb.cpython-36.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/roi_data_layer/__pycache__/roidb.cpython-36.pyc -------------------------------------------------------------------------------- /lib/roi_data_layer/minibatch.py: -------------------------------------------------------------------------------- 1 | # -------------------------------------------------------- 2 | # Fast R-CNN 3 | # Copyright (c) 2015 Microsoft 4 | # Licensed under The MIT License [see LICENSE for details] 5 | # Written by Ross Girshick and Xinlei Chen 6 | # -------------------------------------------------------- 7 | 8 | """Compute minibatch blobs for training a Fast R-CNN network.""" 9 | from __future__ import absolute_import 10 | from __future__ import division 11 | from __future__ import print_function 12 | 13 | import numpy as np 14 | import numpy.random as npr 15 | from scipy.misc import imread 16 | from model.utils.config import cfg 17 | from model.utils.blob import prep_im_for_blob, im_list_to_blob 18 | import pdb 19 | def get_minibatch(roidb, num_classes): 20 | """Given a roidb, construct a minibatch sampled from it.""" 21 | num_images = len(roidb) 22 | # Sample random scales to use for each image in this batch 23 | random_scale_inds = npr.randint(0, high=len(cfg.TRAIN.SCALES), 24 | size=num_images) 25 | assert(cfg.TRAIN.BATCH_SIZE % num_images == 0), \ 26 | 'num_images ({}) must divide BATCH_SIZE ({})'. \ 27 | format(num_images, cfg.TRAIN.BATCH_SIZE) 28 | 29 | # Get the input image blob, formatted for caffe 30 | im_blob, im_scales = _get_image_blob(roidb, random_scale_inds) 31 | 32 | blobs = {'data': im_blob} 33 | 34 | assert len(im_scales) == 1, "Single batch only" 35 | assert len(roidb) == 1, "Single batch only" 36 | 37 | # gt boxes: (x1, y1, x2, y2, cls) 38 | if cfg.TRAIN.USE_ALL_GT: 39 | # Include all ground truth boxes 40 | gt_inds = np.where(roidb[0]['gt_classes'] != 0)[0] 41 | else: 42 | # For the COCO ground truth boxes, exclude the ones that are ''iscrowd'' 43 | gt_inds = np.where(roidb[0]['gt_classes'] != 0 & np.all(roidb[0]['gt_overlaps'].toarray() > -1.0, axis=1))[0] 44 | gt_boxes = np.empty((len(gt_inds), 5), dtype=np.float32) 45 | gt_boxes[:, 0:4] = roidb[0]['boxes'][gt_inds, :] * im_scales[0] 46 | gt_boxes[:, 4] = roidb[0]['gt_classes'][gt_inds] 47 | blobs['gt_boxes'] = gt_boxes 48 | blobs['im_info'] = np.array( 49 | [[im_blob.shape[1], im_blob.shape[2], im_scales[0]]], 50 | dtype=np.float32) 51 | 52 | blobs['img_id'] = roidb[0]['img_id'] 53 | 54 | return blobs 55 | 56 | def _get_image_blob(roidb, scale_inds): 57 | """Builds an input blob from the images in the roidb at the specified 58 | scales. 59 | """ 60 | num_images = len(roidb) 61 | 62 | processed_ims = [] 63 | im_scales = [] 64 | for i in range(num_images): 65 | #im = cv2.imread(roidb[i]['image']) 66 | im = imread(roidb[i]['image']) 67 | 68 | if len(im.shape) == 2: 69 | im = im[:,:,np.newaxis] 70 | im = np.concatenate((im,im,im), axis=2) 71 | # flip the channel, since the original one using cv2 72 | # rgb -> bgr 73 | im = im[:,:,::-1] 74 | 75 | if roidb[i]['flipped']: 76 | im = im[:, ::-1, :] 77 | target_size = cfg.TRAIN.SCALES[scale_inds[i]] 78 | im, im_scale = prep_im_for_blob(im, cfg.PIXEL_MEANS, target_size, 79 | cfg.TRAIN.MAX_SIZE) 80 | im_scales.append(im_scale) 81 | processed_ims.append(im) 82 | 83 | # Create a blob to hold the input images 84 | blob = im_list_to_blob(processed_ims) 85 | 86 | return blob, im_scales 87 | -------------------------------------------------------------------------------- /lib/roi_data_layer/minibatch.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/roi_data_layer/minibatch.pyc -------------------------------------------------------------------------------- /lib/roi_data_layer/roibatchLoader.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/roi_data_layer/roibatchLoader.pyc -------------------------------------------------------------------------------- /lib/roi_data_layer/roidb.py: -------------------------------------------------------------------------------- 1 | """Transform a roidb into a trainable roidb by adding a bunch of metadata.""" 2 | from __future__ import absolute_import 3 | from __future__ import division 4 | from __future__ import print_function 5 | 6 | import datasets 7 | import numpy as np 8 | from model.utils.config import cfg 9 | from datasets.factory import get_imdb 10 | import PIL 11 | import pdb 12 | 13 | def prepare_roidb(imdb): 14 | """Enrich the imdb's roidb by adding some derived quantities that 15 | are useful for training. This function precomputes the maximum 16 | overlap, taken over ground-truth boxes, between each ROI and 17 | each ground-truth box. The class with maximum overlap is also 18 | recorded. 19 | """ 20 | 21 | roidb = imdb.roidb 22 | if not (imdb.name.startswith('coco')): 23 | sizes = [PIL.Image.open(imdb.image_path_at(i)).size 24 | for i in range(imdb.num_images)] 25 | 26 | for i in range(len(imdb.image_index)): 27 | roidb[i]['img_id'] = imdb.image_id_at(i) 28 | roidb[i]['image'] = imdb.image_path_at(i) 29 | if not (imdb.name.startswith('coco')): 30 | roidb[i]['width'] = sizes[i][0] 31 | roidb[i]['height'] = sizes[i][1] 32 | # need gt_overlaps as a dense array for argmax 33 | gt_overlaps = roidb[i]['gt_overlaps'].toarray() 34 | # max overlap with gt over classes (columns) 35 | max_overlaps = gt_overlaps.max(axis=1) 36 | # gt class that had the max overlap 37 | max_classes = gt_overlaps.argmax(axis=1) 38 | roidb[i]['max_classes'] = max_classes 39 | roidb[i]['max_overlaps'] = max_overlaps 40 | # sanity checks 41 | # max overlap of 0 => class should be zero (background) 42 | zero_inds = np.where(max_overlaps == 0)[0] 43 | assert all(max_classes[zero_inds] == 0) 44 | # max overlap > 0 => class should not be zero (must be a fg class) 45 | nonzero_inds = np.where(max_overlaps > 0)[0] 46 | assert all(max_classes[nonzero_inds] != 0) 47 | 48 | 49 | def rank_roidb_ratio(roidb): 50 | # rank roidb based on the ratio between width and height. 51 | ratio_large = 2 # largest ratio to preserve. 52 | ratio_small = 0.5 # smallest ratio to preserve. 53 | 54 | ratio_list = [] 55 | for i in range(len(roidb)): 56 | width = roidb[i]['width'] 57 | height = roidb[i]['height'] 58 | ratio = width / float(height) 59 | 60 | if ratio > ratio_large: 61 | roidb[i]['need_crop'] = 1 62 | ratio = ratio_large 63 | elif ratio < ratio_small: 64 | roidb[i]['need_crop'] = 1 65 | ratio = ratio_small 66 | else: 67 | roidb[i]['need_crop'] = 0 68 | 69 | ratio_list.append(ratio) 70 | 71 | ratio_list = np.array(ratio_list) 72 | ratio_index = np.argsort(ratio_list) 73 | return ratio_list[ratio_index], ratio_index 74 | 75 | def filter_roidb(roidb): 76 | # filter the image without bounding box. 77 | print('before filtering, there are %d images...' % (len(roidb))) 78 | i = 0 79 | while i < len(roidb): 80 | if len(roidb[i]['boxes']) == 0: 81 | del roidb[i] 82 | i -= 1 83 | i += 1 84 | 85 | print('after filtering, there are %d images...' % (len(roidb))) 86 | return roidb 87 | 88 | def combined_roidb(imdb_names, training=True): 89 | """ 90 | Combine multiple roidbs 91 | """ 92 | 93 | def get_training_roidb(imdb): 94 | """Returns a roidb (Region of Interest database) for use in training.""" 95 | if cfg.TRAIN.USE_FLIPPED: 96 | print('Appending horizontally-flipped training examples...') 97 | imdb.append_flipped_images() 98 | print('done') 99 | 100 | print('Preparing training data...') 101 | 102 | prepare_roidb(imdb) 103 | #ratio_index = rank_roidb_ratio(imdb) 104 | print('done') 105 | 106 | return imdb.roidb 107 | 108 | def get_roidb(imdb_name): 109 | imdb = get_imdb(imdb_name) 110 | print('Loaded dataset `{:s}` for training'.format(imdb.name)) 111 | imdb.set_proposal_method(cfg.TRAIN.PROPOSAL_METHOD) 112 | print('Set proposal method: {:s}'.format(cfg.TRAIN.PROPOSAL_METHOD)) 113 | roidb = get_training_roidb(imdb) 114 | return roidb 115 | 116 | roidbs = [get_roidb(s) for s in imdb_names.split('+')] 117 | roidb = roidbs[0] 118 | 119 | if len(roidbs) > 1: 120 | for r in roidbs[1:]: 121 | roidb.extend(r) 122 | tmp = get_imdb(imdb_names.split('+')[1]) 123 | imdb = datasets.imdb.imdb(imdb_names, tmp.classes) 124 | else: 125 | imdb = get_imdb(imdb_names) 126 | 127 | if training: 128 | roidb = filter_roidb(roidb) 129 | 130 | ratio_list, ratio_index = rank_roidb_ratio(roidb) 131 | 132 | return imdb, roidb, ratio_list, ratio_index 133 | -------------------------------------------------------------------------------- /lib/roi_data_layer/roidb.pyc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/lib/roi_data_layer/roidb.pyc -------------------------------------------------------------------------------- /lib/setup.py: -------------------------------------------------------------------------------- 1 | from __future__ import print_function 2 | # -------------------------------------------------------- 3 | # Fast R-CNN 4 | # Copyright (c) 2015 Microsoft 5 | # Licensed under The MIT License [see LICENSE for details] 6 | # Written by Ross Girshick 7 | # -------------------------------------------------------- 8 | 9 | import os 10 | from os.path import join as pjoin 11 | import numpy as np 12 | from distutils.core import setup 13 | from distutils.extension import Extension 14 | from Cython.Distutils import build_ext 15 | 16 | 17 | def find_in_path(name, path): 18 | "Find a file in a search path" 19 | # adapted fom http://code.activestate.com/recipes/52224-find-a-file-given-a-search-path/ 20 | for dir in path.split(os.pathsep): 21 | binpath = pjoin(dir, name) 22 | if os.path.exists(binpath): 23 | return os.path.abspath(binpath) 24 | return None 25 | 26 | 27 | # def locate_cuda(): 28 | # """Locate the CUDA environment on the system 29 | # 30 | # Returns a dict with keys 'home', 'nvcc', 'include', and 'lib64' 31 | # and values giving the absolute path to each directory. 32 | # 33 | # Starts by looking for the CUDAHOME env variable. If not found, everything 34 | # is based on finding 'nvcc' in the PATH. 35 | # """ 36 | # 37 | # # first check if the CUDAHOME env variable is in use 38 | # if 'CUDAHOME' in os.environ: 39 | # home = os.environ['CUDAHOME'] 40 | # nvcc = pjoin(home, 'bin', 'nvcc') 41 | # else: 42 | # # otherwise, search the PATH for NVCC 43 | # default_path = pjoin(os.sep, 'usr', 'local', 'cuda', 'bin') 44 | # nvcc = find_in_path('nvcc', os.environ['PATH'] + os.pathsep + default_path) 45 | # if nvcc is None: 46 | # raise EnvironmentError('The nvcc binary could not be ' 47 | # 'located in your $PATH. Either add it to your path, or set $CUDAHOME') 48 | # home = os.path.dirname(os.path.dirname(nvcc)) 49 | # 50 | # cudaconfig = {'home': home, 'nvcc': nvcc, 51 | # 'include': pjoin(home, 'include'), 52 | # 'lib64': pjoin(home, 'lib64')} 53 | # for k, v in cudaconfig.iteritems(): 54 | # if not os.path.exists(v): 55 | # raise EnvironmentError('The CUDA %s path could not be located in %s' % (k, v)) 56 | # 57 | # return cudaconfig 58 | 59 | 60 | # CUDA = locate_cuda() 61 | 62 | # Obtain the numpy include directory. This logic works across numpy versions. 63 | try: 64 | numpy_include = np.get_include() 65 | except AttributeError: 66 | numpy_include = np.get_numpy_include() 67 | 68 | 69 | def customize_compiler_for_nvcc(self): 70 | """inject deep into distutils to customize how the dispatch 71 | to gcc/nvcc works. 72 | 73 | If you subclass UnixCCompiler, it's not trivial to get your subclass 74 | injected in, and still have the right customizations (i.e. 75 | distutils.sysconfig.customize_compiler) run on it. So instead of going 76 | the OO route, I have this. Note, it's kindof like a wierd functional 77 | subclassing going on.""" 78 | 79 | # tell the compiler it can processes .cu 80 | self.src_extensions.append('.cu') 81 | 82 | # save references to the default compiler_so and _comple methods 83 | default_compiler_so = self.compiler_so 84 | super = self._compile 85 | 86 | # now redefine the _compile method. This gets executed for each 87 | # object but distutils doesn't have the ability to change compilers 88 | # based on source extension: we add it. 89 | def _compile(obj, src, ext, cc_args, extra_postargs, pp_opts): 90 | print(extra_postargs) 91 | if os.path.splitext(src)[1] == '.cu': 92 | # use the cuda for .cu files 93 | self.set_executable('compiler_so', CUDA['nvcc']) 94 | # use only a subset of the extra_postargs, which are 1-1 translated 95 | # from the extra_compile_args in the Extension class 96 | postargs = extra_postargs['nvcc'] 97 | else: 98 | postargs = extra_postargs['gcc'] 99 | 100 | super(obj, src, ext, cc_args, postargs, pp_opts) 101 | # reset the default compiler_so, which we might have changed for cuda 102 | self.compiler_so = default_compiler_so 103 | 104 | # inject our redefined _compile method into the class 105 | self._compile = _compile 106 | 107 | 108 | # run the customize_compiler 109 | class custom_build_ext(build_ext): 110 | def build_extensions(self): 111 | customize_compiler_for_nvcc(self.compiler) 112 | build_ext.build_extensions(self) 113 | 114 | 115 | ext_modules = [ 116 | Extension( 117 | "model.utils.cython_bbox", 118 | ["model/utils/bbox.pyx"], 119 | extra_compile_args={'gcc': ["-Wno-cpp", "-Wno-unused-function"]}, 120 | include_dirs=[numpy_include] 121 | ), 122 | Extension( 123 | 'pycocotools._mask', 124 | sources=['pycocotools/maskApi.c', 'pycocotools/_mask.pyx'], 125 | include_dirs=[numpy_include, 'pycocotools'], 126 | extra_compile_args={ 127 | 'gcc': ['-Wno-cpp', '-Wno-unused-function', '-std=c99']}, 128 | ), 129 | ] 130 | 131 | setup( 132 | name='faster_rcnn', 133 | ext_modules=ext_modules, 134 | # inject our custom trigger 135 | cmdclass={'build_ext': custom_build_ext}, 136 | ) 137 | -------------------------------------------------------------------------------- /lstd_extra_test.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import torchvision 3 | import torchvision.transforms as transforms 4 | 5 | transform = transforms.Compose( 6 | [transforms.ToTensor(), 7 | transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))]) 8 | 9 | trainset = torchvision.datasets.CIFAR10(root='./data', train=True, 10 | download=True, transform=transform) 11 | trainloader = torch.utils.data.DataLoader(trainset, batch_size=4, 12 | shuffle=True, num_workers=2) 13 | 14 | testset = torchvision.datasets.CIFAR10(root='./data', train=False, 15 | download=True, transform=transform) 16 | testloader = torch.utils.data.DataLoader(testset, batch_size=4, 17 | shuffle=False, num_workers=2) 18 | 19 | classes = ('plane', 'car', 'bird', 'cat', 20 | 'deer', 'dog', 'frog', 'horse', 'ship', 'truck') 21 | 22 | import matplotlib.pyplot as plt 23 | import numpy as np 24 | 25 | # functions to show an image 26 | 27 | 28 | def imshow(img): 29 | #print('here') 30 | img = img / 2 + 0.5 # unnormalize 31 | npimg = img.numpy() 32 | plt.cla() 33 | plt.imshow(np.transpose(npimg, (1, 2, 0))) 34 | plt.show() 35 | 36 | # get some random training images 37 | dataiter = iter(trainloader) 38 | images, labels = dataiter.next() 39 | 40 | #imshow(torchvision.utils.make_grid(images)) 41 | #print(' '.join('%5s' % classes[labels[j]] for j in range(4))) 42 | 43 | from torch.autograd import Variable 44 | import torch.nn as nn 45 | import torch.nn.functional as F 46 | 47 | def add_lstd_extras(i): 48 | # Extra layers added to VGG for feature scaling 49 | layers = [] 50 | in_channels = i 51 | conv_add1 = nn.Conv2d(in_channels, 256, 52 | kernel_size=3, stride=1, padding=1) 53 | 54 | in_channels = 256 55 | batchnorm_add1 = nn.BatchNorm2d(in_channels) 56 | conv_add2 = nn.Conv2d(in_channels, 256, 57 | kernel_size=3, stride=2, padding=1) 58 | 59 | batchnorm_add2 = nn.BatchNorm2d(in_channels) 60 | #bbox_score_voc = nn.Linear(256, 21) 61 | 62 | layers += [conv_add1, batchnorm_add1, nn.ReLU(inplace=True), conv_add2, batchnorm_add2, nn.ReLU(inplace=True)] 63 | return layers 64 | 65 | extras_lstd_ = add_lstd_extras(3) 66 | 67 | class Net(nn.Module): 68 | def __init__(self, extras_lstd): 69 | super(Net, self).__init__() 70 | ''' 71 | self.extras_lstd = nn.ModuleList(extras_lstd) 72 | self.classifier = nn.ModuleList([nn.Linear(65536, 10)]) 73 | ''' 74 | ''' 75 | Linear(in_features=25088, out_features=4096, bias=True) 76 | ReLU(inplace) 77 | Dropout(p=0.5) 78 | Linear(in_features=4096, out_features=4096, bias=True) 79 | ReLU(inplace) 80 | Dropout(p=0.5) 81 | 82 | ''' 83 | self.classifier = nn.ModuleList([nn.Linear(3 * 32 * 32, 4096), 84 | nn.ReLU(True), 85 | nn.Dropout(p = 0.5), 86 | nn.Linear(4096, 4096), 87 | nn.ReLU(True), 88 | nn.Dropout(0.5), 89 | nn.Linear(4096, 10)]) 90 | 91 | self.softmax = nn.Softmax(dim=-1) 92 | 93 | 94 | def forward(self, x): 95 | #print(x.size()) 96 | ''' 97 | for k, v in enumerate(self.extras_lstd): 98 | x = v(x) 99 | #print(x.size()) 100 | 101 | x = x.view(x.size(0), -1) 102 | ''' 103 | #print(x.size()) 104 | x = x.view(x.size(0), -1) 105 | for k, v in enumerate(self.classifier): 106 | 107 | x = v(x) 108 | 109 | #cls_output = self.classifier(x) 110 | ''' 111 | x = self.pool(F.relu(self.conv1(x))) 112 | x = self.pool(F.relu(self.conv2(x))) 113 | x = x.view(-1, 16 * 5 * 5) 114 | x = F.relu(self.fc1(x)) 115 | x = F.relu(self.fc2(x)) 116 | x = self.fc3(x) 117 | ''' 118 | return x 119 | 120 | 121 | net = Net(extras_lstd_) 122 | net.cuda() 123 | 124 | import torch.optim as optim 125 | 126 | criterion = nn.CrossEntropyLoss() 127 | optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9) 128 | 129 | for epoch in range(6): # loop over the dataset multiple times 130 | 131 | running_loss = 0.0 132 | for i, data in enumerate(trainloader, 0): 133 | # get the inputs 134 | inputs, labels = data 135 | 136 | # wrap them in Variable 137 | inputs, labels = Variable(inputs.cuda()), Variable(labels.cuda()) 138 | 139 | # zero the parameter gradients 140 | optimizer.zero_grad() 141 | 142 | # forward + backward + optimize 143 | outputs = net(inputs) 144 | loss = criterion(outputs, labels) 145 | loss.backward() 146 | optimizer.step() 147 | 148 | # print statistics 149 | running_loss += loss.data[0] 150 | if i % 2000 == 1999: # print every 2000 mini-batches 151 | print('[%d, %5d] loss: %.3f' % 152 | (epoch + 1, i + 1, running_loss / 2000)) 153 | running_loss = 0.0 154 | 155 | print('Finished Training') 156 | 157 | dataiter = iter(testloader) 158 | images, labels = dataiter.next() 159 | 160 | correct = 0 161 | total = 0 162 | for data in testloader: 163 | images, labels = data 164 | outputs = net(Variable(images.cuda())) 165 | _, predicted = torch.max(outputs.data, 1) 166 | total += labels.size(0) 167 | predicted = predicted.cpu() 168 | correct += (predicted == labels).sum() 169 | 170 | print('Accuracy of the network on the 10000 test images: %d %%' % ( 171 | 100 * correct / total)) 172 | 173 | class_correct = list(0. for i in range(10)) 174 | class_total = list(0. for i in range(10)) 175 | for data in testloader: 176 | images, labels = data 177 | outputs = net(Variable(images.cuda())) 178 | _, predicted = torch.max(outputs.data, 1) 179 | predicted = predicted.cpu() 180 | c = (predicted == labels).squeeze() 181 | for i in range(4): 182 | label = labels[i] 183 | class_correct[label] += c[i] 184 | class_total[label] += 1 185 | 186 | 187 | for i in range(10): 188 | print('Accuracy of %5s : %2d %%' % ( 189 | classes[i], 100 * class_correct[i] / class_total[i])) 190 | -------------------------------------------------------------------------------- /test.py: -------------------------------------------------------------------------------- 1 | from __future__ import print_function 2 | import sys 3 | import os 4 | import argparse 5 | import torch 6 | import torch.nn as nn 7 | import torch.backends.cudnn as cudnn 8 | import torchvision.transforms as transforms 9 | from torch.autograd import Variable 10 | from data import VOCroot, VOC_CLASSES as labelmap 11 | from PIL import Image 12 | from data import AnnotationTransform, VOCDetection, BaseTransform, VOC_CLASSES 13 | import torch.utils.data as data 14 | from ssd import build_ssd 15 | 16 | parser = argparse.ArgumentParser(description='Single Shot MultiBox Detection') 17 | parser.add_argument('--trained_model', default='weights/ssd_300_VOC0712.pth', 18 | type=str, help='Trained state_dict file path to open') 19 | parser.add_argument('--save_folder', default='eval/', type=str, 20 | help='Dir to save results') 21 | parser.add_argument('--visual_threshold', default=0.6, type=float, 22 | help='Final confidence threshold') 23 | parser.add_argument('--cuda', default=False, type=bool, 24 | help='Use cuda to train model') 25 | parser.add_argument('--voc_root', default=VOCroot, help='Location of VOC root directory') 26 | 27 | args = parser.parse_args() 28 | 29 | if not os.path.exists(args.save_folder): 30 | os.mkdir(args.save_folder) 31 | 32 | 33 | def test_net(save_folder, net, cuda, testset, transform, thresh): 34 | # dump predictions and assoc. ground truth to text file for now 35 | filename = save_folder+'test1.txt' 36 | num_images = len(testset) 37 | for i in range(num_images): 38 | print('Testing image {:d}/{:d}....'.format(i+1, num_images)) 39 | img = testset.pull_image(i) 40 | img_id, annotation = testset.pull_anno(i) 41 | x = torch.from_numpy(transform(img)[0]).permute(2, 0, 1) 42 | x = Variable(x.unsqueeze(0)) 43 | 44 | with open(filename, mode='a') as f: 45 | f.write('\nGROUND TRUTH FOR: '+img_id+'\n') 46 | for box in annotation: 47 | f.write('label: '+' || '.join(str(b) for b in box)+'\n') 48 | if cuda: 49 | x = x.cuda() 50 | 51 | y = net(x) # forward pass 52 | detections = y.data 53 | # scale each detection back up to the image 54 | scale = torch.Tensor([img.shape[1], img.shape[0], 55 | img.shape[1], img.shape[0]]) 56 | pred_num = 0 57 | for i in range(detections.size(1)): 58 | j = 0 59 | while detections[0, i, j, 0] >= 0.6: 60 | if pred_num == 0: 61 | with open(filename, mode='a') as f: 62 | f.write('PREDICTIONS: '+'\n') 63 | score = detections[0, i, j, 0] 64 | label_name = labelmap[i-1] 65 | pt = (detections[0, i, j, 1:]*scale).cpu().numpy() 66 | coords = (pt[0], pt[1], pt[2], pt[3]) 67 | pred_num += 1 68 | with open(filename, mode='a') as f: 69 | f.write(str(pred_num)+' label: '+label_name+' score: ' + 70 | str(score) + ' '+' || '.join(str(c) for c in coords) + '\n') 71 | j += 1 72 | 73 | 74 | if __name__ == '__main__': 75 | # load net 76 | num_classes = len(VOC_CLASSES) + 1 # +1 background 77 | net = build_ssd('test', 300, num_classes) # initialize SSD 78 | net.load_state_dict(torch.load(args.trained_model)) 79 | net.eval() 80 | print('Finished loading model!') 81 | # load data 82 | testset = VOCDetection(args.voc_root, [('2007', 'test')], None, AnnotationTransform()) 83 | if args.cuda: 84 | net = net.cuda() 85 | cudnn.benchmark = True 86 | # evaluation 87 | test_net(args.save_folder, net, args.cuda, testset, 88 | BaseTransform(net.size, (104, 117, 123)), 89 | thresh=args.visual_threshold) 90 | -------------------------------------------------------------------------------- /test_overlap.py: -------------------------------------------------------------------------------- 1 | # -*- coding: utf-8 -*- 2 | import torch 3 | import cv2 4 | 5 | def intersect(box_a, box_b): 6 | """ We resize both tensors to [A,B,2] without new malloc: 7 | [A,2] -> [A,1,2] -> [A,B,2] 8 | [B,2] -> [1,B,2] -> [A,B,2] 9 | Then we compute the area of intersect between box_a and box_b. 10 | Args: 11 | box_a: (tensor) bounding boxes, Shape: [A,4]. 12 | box_b: (tensor) bounding boxes, Shape: [B,4]. 13 | Return: 14 | (tensor) intersection area, Shape: [A,B]. 15 | """ 16 | A = box_a.size(0) 17 | B = box_b.size(0) 18 | max_xy = torch.min(box_a[:, 2:].unsqueeze(1).expand(A, B, 2), 19 | box_b[:, 2:].unsqueeze(0).expand(A, B, 2)) 20 | min_xy = torch.max(box_a[:, :2].unsqueeze(1).expand(A, B, 2), 21 | box_b[:, :2].unsqueeze(0).expand(A, B, 2)) 22 | inter = torch.clamp((max_xy - min_xy), min=0) 23 | return inter[:, :, 0] * inter[:, :, 1] 24 | 25 | 26 | def jaccard(box_a, box_b): 27 | """Compute the jaccard overlap of two sets of boxes. The jaccard overlap 28 | is simply the intersection over union of two boxes. Here we operate on 29 | ground truth boxes and default boxes. 30 | E.g.: 31 | A ∩ B / A ∪ B = A ∩ B / (area(A) + area(B) - A ∩ B) 32 | Args: 33 | box_a: (tensor) Ground truth bounding boxes, Shape: [num_objects,4] 34 | box_b: (tensor) Prior boxes from priorbox layers, Shape: [num_priors,4] 35 | Return: 36 | jaccard overlap: (tensor) Shape: [box_a.size(0), box_b.size(0)] 37 | """ 38 | inter = intersect(box_a, box_b) 39 | area_a = ((box_a[:, 2]-box_a[:, 0]) * 40 | (box_a[:, 3]-box_a[:, 1])).unsqueeze(1).expand_as(inter) # [A,B] 41 | area_b = ((box_b[:, 2]-box_b[:, 0]) * 42 | (box_b[:, 3]-box_b[:, 1])).unsqueeze(0).expand_as(inter) # [A,B] 43 | union = area_a + area_b - inter 44 | return inter / union # [A,B] 45 | 46 | 47 | def bbox_overlaps(boxes,query_boxes): 48 | """ 49 | Parameters 50 | ---------- 51 | boxes: (N, 4) ndarray of float 52 | query_boxes: (K, 4) ndarray of float 53 | Returns 54 | ------- 55 | overlaps: (N, K) ndarray of overlap between boxes and query_boxes 56 | """ 57 | 58 | N = boxes.size(0) 59 | K = query_boxes.size(0) 60 | overlaps = torch.FloatTensor(N, K) 61 | overlaps[:] = 0 62 | 63 | for k in range(K): 64 | box_area = ( 65 | (query_boxes[k, 2]*300 - query_boxes[k, 0]*300 + 1) * 66 | (query_boxes[k, 3]*300 - query_boxes[k, 1]*300 + 1) 67 | ) 68 | for n in range(N): 69 | iw = ( 70 | min(boxes[n, 2]*300, query_boxes[k, 2]*300) - 71 | max(boxes[n, 0]*300, query_boxes[k, 0]*300) + 1 72 | ) 73 | if iw > 0: 74 | ih = ( 75 | min(boxes[n, 3]*300, query_boxes[k, 3]*300) - 76 | max(boxes[n, 1]*300, query_boxes[k, 1]*300) + 1 77 | ) 78 | if ih > 0: 79 | ua = float( 80 | (boxes[n, 2]*300 - boxes[n, 0]*300 + 1) * 81 | (boxes[n, 3]*300 - boxes[n, 1]*300 + 1) + 82 | box_area - iw * ih 83 | ) 84 | overlaps[n, k] = iw * ih / ua 85 | return overlaps.cuda() 86 | 87 | 88 | def match_proposal(truths, the_proposal): 89 | """Match each prior box with the ground truth box of the highest jaccard 90 | overlap, encode the bounding boxes, then return the matched indices 91 | corresponding to both confidence and location preds. 92 | Args: 93 | threshold: (float) The overlap threshold used when mathing boxes. 94 | truths: (tensor) Ground truth boxes, Shape: [num_obj, num_priors]. 95 | priors: (tensor) Prior boxes from priorbox layers, Shape: [n_priors,4]. 96 | variances: (tensor) Variances corresponding to each prior coord, 97 | Shape: [num_priors, 4]. 98 | labels: (tensor) All the class labels for the image, Shape: [num_obj]. 99 | loc_t: (tensor) Tensor to be filled w/ endcoded location targets. 100 | conf_t: (tensor) Tensor to be filled w/ matched indices for conf preds. 101 | idx: (int) current batch index 102 | Return: 103 | The matched indices corresponding to 1)location and 2)confidence preds. 104 | """ 105 | 106 | 107 | 108 | overlaps1 = bbox_overlaps( 109 | truths, 110 | the_proposal 111 | ) 112 | overlaps = jaccard( 113 | truths, 114 | the_proposal 115 | ) 116 | best_prior_overlap, best_prior_idx = overlaps.max(1, keepdim=True) 117 | print(best_prior_idx) 118 | # [1,num_priors] best ground truth for each prior 119 | best_truth_overlap, best_truth_idx = overlaps.max(0, keepdim=True) 120 | print(best_truth_idx) 121 | best_truth_idx.squeeze_(0) 122 | print(best_truth_idx) 123 | best_truth_overlap.squeeze_(0) 124 | print(best_truth_overlap) 125 | best_prior_idx.squeeze_(1) 126 | print(best_prior_idx) 127 | best_prior_overlap.squeeze_(1) 128 | print(best_prior_overlap) 129 | best_truth_overlap.index_fill_(0, best_prior_idx, 2) 130 | print(best_truth_overlap) 131 | 132 | print(best_truth_idx) 133 | for j in range(best_prior_idx.size(0)): 134 | best_truth_idx[best_prior_idx[j]] = j 135 | print(best_truth_idx) 136 | return overlaps, overlaps1 137 | 138 | def clamp(boxes): 139 | 140 | for i in range(boxes.size(0)): 141 | box = boxes[i] 142 | for j in range(4): 143 | if box[j] < 0: 144 | box[j] *= -1 145 | if box[2] <= box[0]: 146 | box[2] += box[0] 147 | if box[3] <= box[1]: 148 | box[3] += box[1] 149 | 150 | box[:] /= box.max() 151 | box[:] *= 300 152 | 153 | if __name__ == "__main__": 154 | 155 | truth = torch.randn(3,4) 156 | the_proposal = torch.randn(15,4) 157 | clamp(truth) 158 | clamp(the_proposal) 159 | 160 | print(truth) 161 | print(the_proposal) 162 | 163 | a,b = match_proposal(truth, the_proposal) 164 | 165 | print(a) 166 | print('----------------------') 167 | print(b) -------------------------------------------------------------------------------- /utils/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/JiasiWang/pytorch-lstd/c43b70ee98432599b6d4ba2772132263624603ee/utils/__init__.py --------------------------------------------------------------------------------