├── plant_counting.png ├── requirements.txt ├── snapshots ├── mtc │ └── tasselnetv2plus_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500 │ │ ├── learning_curves.png │ │ ├── model_best.pth.tar │ │ └── tasselnetv2plus_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500.txt ├── wec │ └── tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs8_epoch500 │ │ ├── learning_curves.png │ │ ├── model_best.pth.tar │ │ └── tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs8_epoch500.txt └── shc │ ├── tasselnetv2plus_dataset1_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500 │ ├── learning_curves.png │ ├── model_best.pth.tar │ └── model_ckpt.pth.tar │ └── tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500 │ ├── learning_curves.png │ ├── model_best.pth.tar │ └── tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500.txt ├── config ├── hl_mtc_train.sh ├── hl_wec_train.sh ├── hl_shc_train.sh ├── hl_mtc_eval.sh ├── hl_wec_eval.sh └── hl_shc_eval.sh ├── gen_trainval_list.py ├── data ├── sorghum_head_counting_dataset │ ├── dataset2_train.txt │ ├── dataset2_test.txt │ ├── dataset1_test.txt │ └── dataset1_train.txt ├── wheat_ears_counting_dataset │ ├── val.txt │ └── train.txt └── maize_counting_dataset │ ├── test.txt │ └── train.txt ├── utils.py ├── README.md ├── hlnet.py ├── hldataset.py ├── hltrainval.py └── LICENSE /plant_counting.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/poppinace/tasselnetv2plus/HEAD/plant_counting.png -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- 1 | torch 2 | torchvision 3 | opencv-python 4 | numpy 5 | PIL 6 | matplotlib 7 | scipy 8 | skimage 9 | sklearn 10 | math 11 | json 12 | pandas 13 | h5py 14 | xml -------------------------------------------------------------------------------- /snapshots/mtc/tasselnetv2plus_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500/learning_curves.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/poppinace/tasselnetv2plus/HEAD/snapshots/mtc/tasselnetv2plus_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500/learning_curves.png -------------------------------------------------------------------------------- /snapshots/mtc/tasselnetv2plus_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500/model_best.pth.tar: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/poppinace/tasselnetv2plus/HEAD/snapshots/mtc/tasselnetv2plus_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500/model_best.pth.tar -------------------------------------------------------------------------------- /snapshots/wec/tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs8_epoch500/learning_curves.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/poppinace/tasselnetv2plus/HEAD/snapshots/wec/tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs8_epoch500/learning_curves.png -------------------------------------------------------------------------------- /snapshots/wec/tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs8_epoch500/model_best.pth.tar: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/poppinace/tasselnetv2plus/HEAD/snapshots/wec/tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs8_epoch500/model_best.pth.tar -------------------------------------------------------------------------------- /snapshots/shc/tasselnetv2plus_dataset1_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/learning_curves.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/poppinace/tasselnetv2plus/HEAD/snapshots/shc/tasselnetv2plus_dataset1_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/learning_curves.png -------------------------------------------------------------------------------- /snapshots/shc/tasselnetv2plus_dataset1_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/model_best.pth.tar: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/poppinace/tasselnetv2plus/HEAD/snapshots/shc/tasselnetv2plus_dataset1_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/model_best.pth.tar -------------------------------------------------------------------------------- /snapshots/shc/tasselnetv2plus_dataset1_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/model_ckpt.pth.tar: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/poppinace/tasselnetv2plus/HEAD/snapshots/shc/tasselnetv2plus_dataset1_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/model_ckpt.pth.tar -------------------------------------------------------------------------------- /snapshots/shc/tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/learning_curves.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/poppinace/tasselnetv2plus/HEAD/snapshots/shc/tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/learning_curves.png -------------------------------------------------------------------------------- /snapshots/shc/tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/model_best.pth.tar: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/poppinace/tasselnetv2plus/HEAD/snapshots/shc/tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/model_best.pth.tar -------------------------------------------------------------------------------- /config/hl_mtc_train.sh: -------------------------------------------------------------------------------- 1 | CUDA_VISIBLE_DEVICES=0 python hltrainval.py \ 2 | --data-dir ./data/maize_counting_dataset \ 3 | --dataset mtc \ 4 | --model tasselnetv2plus \ 5 | --exp tasselnetv2plus \ 6 | --data-list ./data/maize_counting_dataset/train.txt \ 7 | --data-val-list ./data/maize_counting_dataset/test.txt \ 8 | --restore-from model_best.pth.tar \ 9 | --image-mean 0.3859 0.4905 0.2895 \ 10 | --image-std 0.1718 0.1712 0.1518 \ 11 | --input-size 64 \ 12 | --output-stride 8 \ 13 | --resize-ratio 0.125 \ 14 | --optimizer sgd \ 15 | --milestones 200 400 \ 16 | --batch-size 9 \ 17 | --crop-size 256 256 \ 18 | --learning-rate 1e-2 \ 19 | --num-epochs 500 \ 20 | --num-workers 0 \ 21 | --print-every 10 \ 22 | --random-seed 2020 \ 23 | --val-every 10 -------------------------------------------------------------------------------- /config/hl_wec_train.sh: -------------------------------------------------------------------------------- 1 | CUDA_VISIBLE_DEVICES=0 python hltrainval.py \ 2 | --data-dir ./data/wheat_ears_counting_dataset \ 3 | --dataset wec \ 4 | --model tasselnetv2plus \ 5 | --exp tasselnetv2plus \ 6 | --data-list ./data/wheat_ears_counting_dataset/train.txt \ 7 | --data-val-list ./data/wheat_ears_counting_dataset/val.txt \ 8 | --restore-from model_best.pth.tar \ 9 | --image-mean 0.4051 0.4392 0.2344 \ 10 | --image-std 0.2569 0.2620 0.2287 \ 11 | --input-size 64 \ 12 | --output-stride 8 \ 13 | --resize-ratio 0.167 \ 14 | --optimizer sgd \ 15 | --milestones 200 400 \ 16 | --batch-size 8 \ 17 | --crop-size 512 512 \ 18 | --learning-rate 1e-2 \ 19 | --num-epochs 500 \ 20 | --num-workers 0 \ 21 | --print-every 10 \ 22 | --random-seed 2020 \ 23 | --val-every 10 -------------------------------------------------------------------------------- /config/hl_shc_train.sh: -------------------------------------------------------------------------------- 1 | CUDA_VISIBLE_DEVICES=0 python hltrainval.py \ 2 | --data-dir ./data/sorghum_head_counting_dataset \ 3 | --dataset shc \ 4 | --model tasselnetv2plus \ 5 | --exp tasselnetv2plus \ 6 | --data-list ./data/sorghum_head_counting_dataset/dataset1_train.txt \ 7 | --data-val-list ./data/sorghum_head_counting_dataset/dataset1_test.txt \ 8 | --restore-from model_best.pth.tar \ 9 | --image-mean 0.3714 0.3609 0.2386 \ 10 | --image-std 0.2705 0.2567 0.2161 \ 11 | --input-size 64 \ 12 | --output-stride 8 \ 13 | --resize-ratio 1 \ 14 | --optimizer sgd \ 15 | --milestones 200 400 \ 16 | --batch-size 5 \ 17 | --crop-size 256 1024 \ 18 | --learning-rate 1e-2 \ 19 | --num-epochs 500 \ 20 | --num-workers 0 \ 21 | --print-every 5 \ 22 | --random-seed 2020 \ 23 | --val-every 10 -------------------------------------------------------------------------------- /config/hl_mtc_eval.sh: -------------------------------------------------------------------------------- 1 | CUDA_VISIBLE_DEVICES=0 python hltrainval.py \ 2 | --data-dir ./data/maize_counting_dataset \ 3 | --dataset mtc \ 4 | --model tasselnetv2plus \ 5 | --exp tasselnetv2plus_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500 \ 6 | --data-list ./data/maize_counting_dataset/train.txt \ 7 | --data-val-list ./data/maize_counting_dataset/test.txt \ 8 | --restore-from model_best.pth.tar \ 9 | --image-mean 0.3859 0.4905 0.2895 \ 10 | --image-std 0.1718 0.1712 0.1518 \ 11 | --input-size 64 \ 12 | --output-stride 8 \ 13 | --resize-ratio 0.125 \ 14 | --optimizer sgd \ 15 | --milestones 200 400 \ 16 | --batch-size 9 \ 17 | --crop-size 256 256 \ 18 | --learning-rate 1e-2 \ 19 | --num-epochs 500 \ 20 | --num-workers 0 \ 21 | --print-every 10 \ 22 | --random-seed 2020 \ 23 | --val-every 10 \ 24 | --evaluate-only \ 25 | --save-output -------------------------------------------------------------------------------- /config/hl_wec_eval.sh: -------------------------------------------------------------------------------- 1 | CUDA_VISIBLE_DEVICES=0 python hltrainval.py \ 2 | --data-dir ./data/wheat_ears_counting_dataset \ 3 | --dataset wec \ 4 | --model tasselnetv2plus \ 5 | --exp tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs8_epoch500 \ 6 | --data-list ./data/wheat_ears_counting_dataset/train.txt \ 7 | --data-val-list ./data/wheat_ears_counting_dataset/val.txt \ 8 | --restore-from model_best.pth.tar \ 9 | --image-mean 0.4051 0.4392 0.2344 \ 10 | --image-std 0.2569 0.2620 0.2287 \ 11 | --input-size 64 \ 12 | --output-stride 8 \ 13 | --resize-ratio 0.167 \ 14 | --optimizer sgd \ 15 | --milestones 200 400 \ 16 | --batch-size 8 \ 17 | --crop-size 512 512 \ 18 | --learning-rate 1e-2 \ 19 | --num-epochs 500 \ 20 | --num-workers 0 \ 21 | --print-every 10 \ 22 | --random-seed 2020 \ 23 | --val-every 10 \ 24 | --evaluate-only \ 25 | --save-output -------------------------------------------------------------------------------- /config/hl_shc_eval.sh: -------------------------------------------------------------------------------- 1 | CUDA_VISIBLE_DEVICES=0 python hltrainval.py \ 2 | --data-dir ./data/sorghum_head_counting_dataset \ 3 | --dataset shc \ 4 | --model tasselnetv2plus \ 5 | --exp tasselnetv2plus_dataset1_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500 \ 6 | --data-list ./data/sorghum_head_counting_dataset/dataset1_train.txt \ 7 | --data-val-list ./data/sorghum_head_counting_dataset/dataset1_test.txt \ 8 | --restore-from model_best.pth.tar \ 9 | --image-mean 0.3714 0.3609 0.2386 \ 10 | --image-std 0.2705 0.2567 0.2161 \ 11 | --input-size 64 \ 12 | --output-stride 8 \ 13 | --resize-ratio 1 \ 14 | --optimizer sgd \ 15 | --milestones 200 400 \ 16 | --batch-size 5 \ 17 | --crop-size 256 1024 \ 18 | --learning-rate 1e-2 \ 19 | --num-epochs 500 \ 20 | --num-workers 0 \ 21 | --print-every 5 \ 22 | --random-seed 2020 \ 23 | --val-every 10 \ 24 | --evaluate-only \ 25 | --save-output -------------------------------------------------------------------------------- /gen_trainval_list.py: -------------------------------------------------------------------------------- 1 | import os 2 | import glob 3 | import random 4 | 5 | root = './data/wheat_ears_counting_dataset' 6 | image_folder = 'images' 7 | label_folder = 'labels' 8 | train = 'train' 9 | val = 'val' 10 | 11 | train_path = os.path.join(root, train) 12 | with open('train.txt', 'w') as f: 13 | for image_path in glob.glob(os.path.join(train_path, image_folder, '*.JPG')): 14 | im_path = image_path.replace(root, '') 15 | gt_path = im_path.replace(image_folder, label_folder).replace('.JPG', '.xml') 16 | f.write(im_path+'\t'+gt_path+'\n') 17 | 18 | val_path = os.path.join(root, val) 19 | with open('val.txt', 'w') as f: 20 | for image_path in glob.glob(os.path.join(val_path, image_folder, '*.JPG')): 21 | im_path = image_path.replace(root, '') 22 | gt_path = im_path.replace(image_folder, label_folder).replace('.JPG', '.xml') 23 | f.write(im_path+'\t'+gt_path+'\n') -------------------------------------------------------------------------------- /data/sorghum_head_counting_dataset/dataset2_train.txt: -------------------------------------------------------------------------------- 1 | /original/dataset2/C1-R20-G18-DSC01050.tif /labeled/dataset2/C1-R20-G18-DSC01050-hand.png 2 | /original/dataset2/C16-R9-G127-DSC00827.tif /labeled/dataset2/C16-R9-G127-DSC00827-hand.png 3 | /original/dataset2/C5-R13-G139-DSC00885.tif /labeled/dataset2/C5-R13-G139-DSC00885-hand.png 4 | /original/dataset2/C5-R25-G148-DSC01295.tif /labeled/dataset2/C5-R25-G148-DSC01295-hand.png 5 | /original/dataset2/C18-R21-G519-DSC01086.tif /labeled/dataset2/C18-R21-G519-DSC01086-hand.png 6 | /original/dataset2/C4-R3-G98-DSC00677.tif /labeled/dataset2/C4-R3-G98-DSC00677-hand.png 7 | /original/dataset2/C21-R10-G471-DSC00815.tif /labeled/dataset2/C21-R10-G471-DSC00815-hand.png 8 | /original/dataset2/C15-R39-G98-DSC01883.tif /labeled/dataset2/C15-R39-G98-DSC01883-hand.png 9 | /original/dataset2/C18-R4-G184-DSC00648.tif /labeled/dataset2/C18-R4-G184-DSC00648-hand.png 10 | /original/dataset2/C11-R13-G325-DSC00897.tif /labeled/dataset2/C11-R13-G325-DSC00897-hand.png 11 | /original/dataset2/C3-R20-G34-DSC01054.tif /labeled/dataset2/C3-R20-G34-DSC01054-hand.png 12 | /original/dataset2/C13-R28-G47-DSC01454.tif /labeled/dataset2/C13-R28-G47-DSC01454-hand.png 13 | /original/dataset2/C24-R27-G669-DSC01359.tif /labeled/dataset2/C24-R27-G669-DSC01359-hand.png 14 | /original/dataset2/C2-R1-G34-DSC00532.tif /labeled/dataset2/C2-R1-G34-DSC00532-hand.png 15 | /original/dataset2/C34-R34-G300-DSC01818.tif /labeled/dataset2/C34-R34-G300-DSC01818-hand.png 16 | /original/dataset2/C20-R21-G139-DSC01090.tif /labeled/dataset2/C20-R21-G139-DSC01090-hand.png 17 | /original/dataset2/C30-R10-G569-DSC00795.tif /labeled/dataset2/C30-R10-G569-DSC00795-hand.png 18 | /original/dataset2/C4-R39-G127-DSC01906.tif /labeled/dataset2/C4-R39-G127-DSC01906-hand.png 19 | /original/dataset2/C35-R17-G899-DSC00960.tif /labeled/dataset2/C35-R17-G899-DSC00960-hand.png 20 | /original/dataset2/C36-R20-G899-DSC01123.tif /labeled/dataset2/C36-R20-G899-DSC01123-hand.png 21 | -------------------------------------------------------------------------------- /data/sorghum_head_counting_dataset/dataset2_test.txt: -------------------------------------------------------------------------------- 1 | /original/dataset2/C13-R14-G384-DSC00901.tif /labeled/dataset2/C13-R14-G384-DSC00901-hand.png 2 | /original/dataset2/C11-R16-G328-DSC01010.tif /labeled/dataset2/C11-R16-G328-DSC01010-hand.png 3 | /original/dataset2/C16-R6-G148-DSC00733.tif /labeled/dataset2/C16-R6-G148-DSC00733-hand.png 4 | /original/dataset2/C20-R20-G569-DSC01090.tif /labeled/dataset2/C20-R20-G569-DSC01090-hand.png 5 | /original/dataset2/C25-R7-G519-DSC00752.tif /labeled/dataset2/C25-R7-G519-DSC00752-hand.png 6 | /original/dataset2/C28-R22-G384-DSC01147.tif /labeled/dataset2/C28-R22-G384-DSC01147-hand.png 7 | /original/dataset2/C8-R17-G240-DSC01017.tif /labeled/dataset2/C8-R17-G240-DSC01017-hand.png 8 | /original/dataset2/C23-R39-G287-DSC01786.tif /labeled/dataset2/C23-R39-G287-DSC01786-hand.png 9 | /original/dataset2/C24-R10-G579-DSC00808.tif /labeled/dataset2/C24-R10-G579-DSC00808-hand.png 10 | /original/dataset2/C20-R38-G579-DSC01780.tif /labeled/dataset2/C20-R38-G579-DSC01780-hand.png 11 | /original/dataset2/C2-R17-G47-DSC01029.tif /labeled/dataset2/C2-R17-G47-DSC01029-hand.png 12 | /original/dataset2/C10-R1-G287-DSC00549.tif /labeled/dataset2/C10-R1-G287-DSC00549-hand.png 13 | /original/dataset2/C6-R28-G184-DSC01469.tif /labeled/dataset2/C6-R28-G184-DSC01469-hand.png 14 | /original/dataset2/C27-R21-G325-DSC01104.tif /labeled/dataset2/C27-R21-G325-DSC01104-hand.png 15 | /original/dataset2/C30-R19-G669-DSC01111.tif /labeled/dataset2/C30-R19-G669-DSC01111-hand.png 16 | /original/dataset2/C25-R22-G240-DSC01154.tif /labeled/dataset2/C25-R22-G240-DSC01154-hand.png 17 | /original/dataset2/C2-R4-G18-DSC00681.tif /labeled/dataset2/C2-R4-G18-DSC00681-hand.png 18 | /original/dataset2/C10-R18-G300-DSC01013.tif /labeled/dataset2/C10-R18-G300-DSC01013-hand.png 19 | /original/dataset2/C33-R39-G328-DSC01845.tif /labeled/dataset2/C33-R39-G328-DSC01845-hand.png 20 | /original/dataset2/C16-R22-G471-DSC01173.tif /labeled/dataset2/C16-R22-G471-DSC01173-hand.png 21 | -------------------------------------------------------------------------------- /data/sorghum_head_counting_dataset/dataset1_test.txt: -------------------------------------------------------------------------------- 1 | /original/dataset1/DSC01298-crop.jpg /labeled/dataset1/DSC01298-crop-hand.png 2 | /original/dataset1/DSC02083-crop.jpg /labeled/dataset1/DSC02083-crop-hand.png 3 | /original/dataset1/DSC01562-crop.jpg /labeled/dataset1/DSC01562-crop-hand.png 4 | /original/dataset1/DSC02151-crop.jpg /labeled/dataset1/DSC02151-crop-hand.png 5 | /original/dataset1/DSC00732-crop.jpg /labeled/dataset1/DSC00732-crop-hand.png 6 | /original/dataset1/DSC02262-crop.jpg /labeled/dataset1/DSC02262-crop-hand.png 7 | /original/dataset1/DSC00527-crop.jpg /labeled/dataset1/DSC00527-crop-hand.png 8 | /original/dataset1/DSC01338-crop.jpg /labeled/dataset1/DSC01338-crop-hand.png 9 | /original/dataset1/DSC01537-crop.jpg /labeled/dataset1/DSC01537-crop-hand.png 10 | /original/dataset1/DSC00706-crop.jpg /labeled/dataset1/DSC00706-crop-hand.png 11 | /original/dataset1/DSC01777-crop.jpg /labeled/dataset1/DSC01777-crop-hand.png 12 | /original/dataset1/DSC00971-crop.jpg /labeled/dataset1/DSC00971-crop-hand.png 13 | /original/dataset1/DSC00655-crop.jpg /labeled/dataset1/DSC00655-crop-hand.png 14 | /original/dataset1/DSC02085-crop.jpg /labeled/dataset1/DSC02085-crop-hand.png 15 | /original/dataset1/DSC02286-crop.jpg /labeled/dataset1/DSC02286-crop-hand.png 16 | /original/dataset1/DSC02263-crop.jpg /labeled/dataset1/DSC02263-crop-hand.png 17 | /original/dataset1/DSC01807-crop.jpg /labeled/dataset1/DSC01807-crop-hand.png 18 | /original/dataset1/DSC01528-crop.jpg /labeled/dataset1/DSC01528-crop-hand.png 19 | /original/dataset1/DSC00889-crop.jpg /labeled/dataset1/DSC00889-crop-hand.png 20 | /original/dataset1/DSC02011-crop.jpg /labeled/dataset1/DSC02011-crop-hand.png 21 | /original/dataset1/DSC02413-crop.jpg /labeled/dataset1/DSC02413-crop-hand.png 22 | /original/dataset1/DSC02425-crop.jpg /labeled/dataset1/DSC02425-crop-hand.png 23 | /original/dataset1/DSC00895-crop.jpg /labeled/dataset1/DSC00895-crop-hand.png 24 | /original/dataset1/DSC02401-crop.jpg /labeled/dataset1/DSC02401-crop-hand.png 25 | /original/dataset1/DSC02562-crop.jpg /labeled/dataset1/DSC02562-crop-hand.png 26 | /original/dataset1/DSC02258-crop.jpg /labeled/dataset1/DSC02258-crop-hand.png 27 | -------------------------------------------------------------------------------- /data/sorghum_head_counting_dataset/dataset1_train.txt: -------------------------------------------------------------------------------- 1 | /original/dataset1/DSC01890-crop.jpg /labeled/dataset1/DSC01890-crop-hand.png 2 | /original/dataset1/DSC00952-crop.jpg /labeled/dataset1/DSC00952-crop-hand.png 3 | /original/dataset1/DSC00554-crop.jpg /labeled/dataset1/DSC00554-crop-hand.png 4 | /original/dataset1/DSC02339-crop.jpg /labeled/dataset1/DSC02339-crop-hand.png 5 | /original/dataset1/DSC00613-crop.jpg /labeled/dataset1/DSC00613-crop-hand.png 6 | /original/dataset1/DSC00941-crop.jpg /labeled/dataset1/DSC00941-crop-hand.png 7 | /original/dataset1/DSC01983-crop.jpg /labeled/dataset1/DSC01983-crop-hand.png 8 | /original/dataset1/DSC02291-crop.jpg /labeled/dataset1/DSC02291-crop-hand.png 9 | /original/dataset1/DSC01154-crop.jpg /labeled/dataset1/DSC01154-crop-hand.png 10 | /original/dataset1/DSC01024-crop.jpg /labeled/dataset1/DSC01024-crop-hand.png 11 | /original/dataset1/DSC00748-crop.jpg /labeled/dataset1/DSC00748-crop-hand.png 12 | /original/dataset1/DSC01402-crop.jpg /labeled/dataset1/DSC01402-crop-hand.png 13 | /original/dataset1/DSC00781-crop.jpg /labeled/dataset1/DSC00781-crop-hand.png 14 | /original/dataset1/DSC00532-crop.jpg /labeled/dataset1/DSC00532-crop-hand.png 15 | /original/dataset1/DSC01826-crop.jpg /labeled/dataset1/DSC01826-crop-hand.png 16 | /original/dataset1/DSC02241-crop.jpg /labeled/dataset1/DSC02241-crop-hand.png 17 | /original/dataset1/DSC01748-crop.jpg /labeled/dataset1/DSC01748-crop-hand.png 18 | /original/dataset1/DSC02441-crop.jpg /labeled/dataset1/DSC02441-crop-hand.png 19 | /original/dataset1/DSC01367-crop.jpg /labeled/dataset1/DSC01367-crop-hand.png 20 | /original/dataset1/DSC02120-crop.jpg /labeled/dataset1/DSC02120-crop-hand.png 21 | /original/dataset1/DSC02084-crop.jpg /labeled/dataset1/DSC02084-crop-hand.png 22 | /original/dataset1/DSC02487-crop.jpg /labeled/dataset1/DSC02487-crop-hand.png 23 | /original/dataset1/DSC01023-crop.jpg /labeled/dataset1/DSC01023-crop-hand.png 24 | /original/dataset1/DSC01288-crop.jpg /labeled/dataset1/DSC01288-crop-hand.png 25 | /original/dataset1/DSC01317-crop.jpg /labeled/dataset1/DSC01317-crop-hand.png 26 | /original/dataset1/DSC01289-crop.jpg /labeled/dataset1/DSC01289-crop-hand.png 27 | -------------------------------------------------------------------------------- /utils.py: -------------------------------------------------------------------------------- 1 | """ 2 | @author: hao 3 | """ 4 | 5 | import scipy 6 | import numpy as np 7 | import math 8 | import cv2 as cv 9 | from scipy.ndimage import gaussian_filter, morphology 10 | from skimage.measure import label, regionprops 11 | from sklearn import linear_model 12 | import matplotlib.pyplot as plt 13 | 14 | 15 | def compute_mae(pd, gt): 16 | pd, gt = np.array(pd), np.array(gt) 17 | diff = pd - gt 18 | mae = np.mean(np.abs(diff)) 19 | return mae 20 | 21 | 22 | def compute_mse(pd, gt): 23 | pd, gt = np.array(pd), np.array(gt) 24 | diff = pd - gt 25 | mse = np.sqrt(np.mean((diff ** 2))) 26 | return mse 27 | 28 | def compute_relerr(pd, gt): 29 | pd, gt = np.array(pd), np.array(gt) 30 | diff = pd - gt 31 | diff = diff[gt > 0] 32 | gt = gt[gt > 0] 33 | if (diff is not None) and (gt is not None): 34 | rmae = np.mean(np.abs(diff) / gt) * 100 35 | rmse = np.sqrt(np.mean(diff**2 / gt**2)) * 100 36 | else: 37 | rmae = 0 38 | rmse = 0 39 | return rmae, rmse 40 | 41 | 42 | def rsquared(pd, gt): 43 | """ Return R^2 where x and y are array-like.""" 44 | pd, gt = np.array(pd), np.array(gt) 45 | slope, intercept, r_value, p_value, std_err = scipy.stats.linregress(pd, gt) 46 | return r_value**2 47 | 48 | 49 | def dense_sample2d(x, sx, stride): 50 | (h,w) = x.shape[:2] 51 | #idx_img = np.array([i for i in range(h*w)]).reshape(h,w) 52 | idx_img = np.zeros((h,w),dtype=float) 53 | 54 | th = [i for i in range(0, h-sx+1, stride)] 55 | tw = [j for j in range(0, w-sx+1, stride)] 56 | norm_vec = np.zeros(len(th)*len(tw)) 57 | 58 | for i in th: 59 | for j in tw: 60 | idx_img[i:i+sx,j:j+sx] = idx_img[i:i+sx,j:j+sx]+1 61 | 62 | # # plot redundancy map 63 | # import os 64 | # import matplotlib.pyplot as plt 65 | # cmap = plt.cm.get_cmap('hot') 66 | # idx_img = idx_img / (idx_img.max()) 67 | # idx_img = cmap(idx_img) * 255. 68 | # plt.figure() 69 | # plt.imshow(idx_img.astype(np.uint8)) 70 | # plt.axis('off') 71 | # plt.savefig(os.path.join('redundancy_map.pdf'), bbox_inches='tight', dpi = 300) 72 | # plt.close() 73 | 74 | idx_img = 1/idx_img 75 | idx_img = idx_img/sx/sx 76 | #line order 77 | idx = 0 78 | for i in th: 79 | for j in tw: 80 | norm_vec[idx] =idx_img[i:i+sx,j:j+sx].sum() 81 | idx+=1 82 | 83 | return norm_vec 84 | 85 | 86 | def recover_countmap(pred, image, patch_sz, stride): 87 | pred = pred.reshape(-1) 88 | imH, imW = image.shape[2:4] 89 | cntMap = np.zeros((imH, imW), dtype=float) 90 | norMap = np.zeros((imH, imW), dtype=float) 91 | 92 | H = np.arange(0, imH - patch_sz + 1, stride) 93 | W = np.arange(0, imW - patch_sz + 1, stride) 94 | cnt = 0 95 | for h in H: 96 | for w in W: 97 | pixel_cnt = pred[cnt] / patch_sz / patch_sz 98 | cntMap[h:h+patch_sz, w:w+patch_sz] += pixel_cnt 99 | norMap[h:h+patch_sz, w:w+patch_sz] += np.ones((patch_sz,patch_sz)) 100 | cnt += 1 101 | return cntMap / (norMap + 1e-12) 102 | -------------------------------------------------------------------------------- /data/wheat_ears_counting_dataset/val.txt: -------------------------------------------------------------------------------- 1 | /val/images/5172.JPG /val/labels/5172.xml 2 | /val/images/1161.JPG /val/labels/1161.xml 3 | /val/images/3192.JPG /val/labels/3192.xml 4 | /val/images/3191.JPG /val/labels/3191.xml 5 | /val/images/4171.JPG /val/labels/4171.xml 6 | /val/images/5182.JPG /val/labels/5182.xml 7 | /val/images/2201.JPG /val/labels/2201.xml 8 | /val/images/1052.JPG /val/labels/1052.xml 9 | /val/images/3201.JPG /val/labels/3201.xml 10 | /val/images/4192.JPG /val/labels/4192.xml 11 | /val/images/1122.JPG /val/labels/1122.xml 12 | /val/images/3151.JPG /val/labels/3151.xml 13 | /val/images/3101.JPG /val/labels/3101.xml 14 | /val/images/6132.JPG /val/labels/6132.xml 15 | /val/images/3162.JPG /val/labels/3162.xml 16 | /val/images/2021.JPG /val/labels/2021.xml 17 | /val/images/5142.JPG /val/labels/5142.xml 18 | /val/images/5092.JPG /val/labels/5092.xml 19 | /val/images/6012.JPG /val/labels/6012.xml 20 | /val/images/6021.JPG /val/labels/6021.xml 21 | /val/images/4122.JPG /val/labels/4122.xml 22 | /val/images/2112.JPG /val/labels/2112.xml 23 | /val/images/3202.JPG /val/labels/3202.xml 24 | /val/images/5061.JPG /val/labels/5061.xml 25 | /val/images/5141.JPG /val/labels/5141.xml 26 | /val/images/3152.JPG /val/labels/3152.xml 27 | /val/images/2072.JPG /val/labels/2072.xml 28 | /val/images/6062.JPG /val/labels/6062.xml 29 | /val/images/4191.JPG /val/labels/4191.xml 30 | /val/images/2052.JPG /val/labels/2052.xml 31 | /val/images/4082.JPG /val/labels/4082.xml 32 | /val/images/4131.JPG /val/labels/4131.xml 33 | /val/images/2051.JPG /val/labels/2051.xml 34 | /val/images/1051.JPG /val/labels/1051.xml 35 | /val/images/1162.JPG /val/labels/1162.xml 36 | /val/images/4081.JPG /val/labels/4081.xml 37 | /val/images/4021.JPG /val/labels/4021.xml 38 | /val/images/4132.JPG /val/labels/4132.xml 39 | /val/images/2071.JPG /val/labels/2071.xml 40 | /val/images/1061.JPG /val/labels/1061.xml 41 | /val/images/1011.JPG /val/labels/1011.xml 42 | /val/images/1062.JPG /val/labels/1062.xml 43 | /val/images/6011.JPG /val/labels/6011.xml 44 | /val/images/5122.JPG /val/labels/5122.xml 45 | /val/images/3161.JPG /val/labels/3161.xml 46 | /val/images/6131.JPG /val/labels/6131.xml 47 | /val/images/5181.JPG /val/labels/5181.xml 48 | /val/images/3102.JPG /val/labels/3102.xml 49 | /val/images/2111.JPG /val/labels/2111.xml 50 | /val/images/1101.JPG /val/labels/1101.xml 51 | /val/images/2102.JPG /val/labels/2102.xml 52 | /val/images/5121.JPG /val/labels/5121.xml 53 | /val/images/6061.JPG /val/labels/6061.xml 54 | /val/images/6161.JPG /val/labels/6161.xml 55 | /val/images/1012.JPG /val/labels/1012.xml 56 | /val/images/5091.JPG /val/labels/5091.xml 57 | /val/images/6052.JPG /val/labels/6052.xml 58 | /val/images/4022.JPG /val/labels/4022.xml 59 | /val/images/5062.JPG /val/labels/5062.xml 60 | /val/images/2202.JPG /val/labels/2202.xml 61 | /val/images/3071.JPG /val/labels/3071.xml 62 | /val/images/1102.JPG /val/labels/1102.xml 63 | /val/images/2022.JPG /val/labels/2022.xml 64 | /val/images/5171.JPG /val/labels/5171.xml 65 | /val/images/6022.JPG /val/labels/6022.xml 66 | /val/images/1121.JPG /val/labels/1121.xml 67 | /val/images/6162.JPG /val/labels/6162.xml 68 | /val/images/4121.JPG /val/labels/4121.xml 69 | /val/images/3072.JPG /val/labels/3072.xml 70 | /val/images/6051.JPG /val/labels/6051.xml 71 | /val/images/4172.JPG /val/labels/4172.xml -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # TasselNetv2+ 2 | 3 |

4 | 5 |

6 | 7 | This repository includes the official implementation of TasselNetv2+ for plant counting, presented in paper: 8 | 9 | **TasselNetv2+: A Fast Implementation for High-Throughput Plant Counting from High-Resolution RGB Imagery** 10 | 11 | Frontiers in Plant Science, 2020 12 | 13 | [Hao Lu](https://sites.google.com/site/poppinace/) and Zhiguo Cao 14 | 15 | 16 | ## Highlights 17 | - **Highly Efficient:** TasselNetv2+ runs an order of magnitude faster than [TasselNetv2](https://link.springer.com/article/10.1186/s13007-019-0537-2) with around 30fps on image resolution of 1980×1080 on a single GTX 1070; 18 | - **Effective:** It retrains the same level of counting accuracy compared to its counterpart TasselNetv2; 19 | - **Easy to Use:** Pretrained plant counting models are included in this repository. 20 | 21 | 22 | ## Installation 23 | The code has been tested on Python 3.7.4 and PyTorch 1.2.0. Please follow the official instructions to configure your environment. See other required packages in `requirements.txt`. 24 | 25 | ## Prepare Your Data 26 | **Wheat Ears Counting** 27 | 1. Download the Wheat Ears Counting (WEC) dataset from: [Google Drive (2.5 GB)](https://drive.google.com/open?id=1XHcTqRWf-xD-WuBeJ0C9KfIN8ye6cnSs). I have reorganized the data, the credit of this dataset belongs to [this repository](https://github.com/simonMadec/Wheat-Ears-Detection-Dataset). 28 | 2. Unzip the dataset and move it into the `./data` folder, the path structure should look like this: 29 | ```` 30 | $./data/wheat_ears_counting_dataset 31 | ├──── train 32 | │ ├──── images 33 | │ └──── labels 34 | ├──── val 35 | │ ├──── images 36 | │ └──── labels 37 | ```` 38 | 39 | **Maize Tassels Counting** 40 | 1. Download the Maize Tassels Counting (MTC) dataset from: [Google Drive (1.8 GB)](https://drive.google.com/open?id=1IyGpYMS_6eClco2zpHKzW5QDUuZqfVFJ) 41 | 2. Unzip the dataset and move it into the `./data` folder, the path structure should look like this: 42 | ```` 43 | $./data/maize_counting_dataset 44 | ├──── trainval 45 | │ ├──── images 46 | │ └──── labels 47 | ├──── test 48 | │ ├──── images 49 | │ └──── labels 50 | ```` 51 | 52 | **Sorghum Heads Counting** 53 | 1. Download the Sorghum Heads Counting (SHC) dataset from: [Google Drive (152 MB)](https://drive.google.com/open?id=1msk8vYDyKdrYDq5zU1kKWOxfmgaXpy-P). The credit of this dataset belongs to [this repository](https://github.com/oceam/sorghum-head). I only use the two subsets that have dotted annotations available. 54 | 2. Unzip the dataset and move it into the `./data` folder, the path structure should look like this: 55 | ```` 56 | $./data/sorghum_head_counting_dataset 57 | ├──── original 58 | │ ├──── dataset1 59 | │ └──── dataset2 60 | ├──── labeled 61 | │ ├──── dataset1 62 | │ └──── dataset2 63 | ```` 64 | 65 | ## Inference 66 | Run the following command to reproduce our results of TasselNetv2+ on the WEC/MTC/SHC dataset: 67 | 68 | sh config/hl_wec_eval.sh 69 | 70 | sh config/hl_mtc_eval.sh 71 | 72 | sh config/hl_shc_eval.sh 73 | 74 | - Results are saved in the path `./results/$dataset/$exp/$epoch`. 75 | 76 | ## Training 77 | Run the following command to train TasselNetv2+ on the on the WEC/MTC/SHC dataset: 78 | 79 | sh config/hl_wec_train.sh 80 | 81 | sh config/hl_mtc_train.sh 82 | 83 | sh config/hl_shc_train.sh 84 | 85 | 86 | ## Play with Your Own Dataset 87 | To use this framework on your own dataset, you may need to: 88 | 1. Annotate your data with dotted annotations. I recommend the [VGG Image Annotator](http://www.robots.ox.ac.uk/~vgg/software/via/); 89 | 2. Generate train/validation list following the example in `gen_trainval_list.py`; 90 | 3. Write your dataloader following example codes in `hldataset.py`; 91 | 4. Compute the mean and standard deviation of RGB on the training set; 92 | 5. Create a new entry in the `dataset_list` in `hltrainval.py`; 93 | 6. Create a new `your_dataset.sh` following examples in `./config` and modify the hyper-parameters (e.g., batch size, crop size) if applicable. 94 | 7. Train and test your model. Happy playing:) 95 | 96 | ## Citation 97 | If you find this work or code useful for your research, please cite: 98 | ``` 99 | @article{lu2020tasselnetv2plus, 100 | title={TasselNetV2+: A fast implementation for high-throughput plant counting from high-resolution RGB imagery}, 101 | author={Lu, Hao and Cao, Zhiguo}, 102 | journal={Frontiers in Plant Science}, 103 | year={2020} 104 | } 105 | 106 | @article{xiong2019tasselnetv2, 107 | title={TasselNetv2: in-field counting of wheat spikes with context-augmented local regression networks}, 108 | author={Xiong, Haipeng and Cao, Zhiguo and Lu, Hao and Madec, Simon and Liu, Liang and Shen, Chunhua}, 109 | journal={Plant Methods}, 110 | volume={15}, 111 | number={1}, 112 | pages={150}, 113 | year={2019}, 114 | publisher={Springer} 115 | } 116 | ``` 117 | 118 | ## Permission 119 | This code is only for non-commercial purposes. Please contact Hao Lu (hlu@hust.edu.cn) if you are interested in commerial use. 120 | -------------------------------------------------------------------------------- /snapshots/wec/tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs8_epoch500/tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs8_epoch500.txt: -------------------------------------------------------------------------------- 1 | image_scale : 0.00392156862745098 2 | image_mean : [[[0.4051 0.4392 0.2344]]] 3 | image_std : [[[0.2569 0.262 0.2287]]] 4 | scales : [0.7, 1, 1.3] 5 | shorter_side : 224 6 | data_dir : ./data/wheat_ears_counting_dataset 7 | dataset : wec 8 | exp : tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs10_epoch500 9 | data_list : ./data/wheat_ears_counting_dataset/train.txt 10 | data_val_list : ./data/wheat_ears_counting_dataset/val.txt 11 | restore_from : ./snapshots/wec/tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs10_epoch500/model_best.pth.tar 12 | snapshot_dir : ./snapshots/wec/tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs10_epoch500 13 | result_dir : ./results/wec/tasselnetv2plus_rf110_i64o8_r0167_crop512_lr-2_bs10_epoch500 14 | save_output : False 15 | input_size : 64 16 | output_stride : 8 17 | resize_ratio : 0.1667 18 | model : tasselnetv2plus 19 | width_mult : 1.0 20 | use_pretrained : False 21 | freeze_bn : False 22 | sync_bn : False 23 | use_nonlinear : False 24 | use_context : False 25 | use_squeeze : False 26 | optimizer : sgd 27 | batch_size : 8 28 | milestones : [200, 400] 29 | crop_size : (512, 512) 30 | evaluate_only : False 31 | learning_rate : 0.01 32 | momentum : 0.95 33 | weight_decay : 0.0005 34 | mult : 1 35 | num_epochs : 500 36 | num_workers : 0 37 | print_every : 10 38 | random_seed : 2020 39 | val_every : 10 40 | epoch: 10, mae: 55.73, mse: 59.27, rmae: 41.13%, rmse: 42.95%, r2: 0.1498 41 | epoch: 20, mae: 22.46, mse: 27.32, rmae: 17.31%, rmse: 21.22%, r2: 0.2067 42 | epoch: 30, mae: 26.42, mse: 30.22, rmae: 19.66%, rmse: 22.12%, r2: 0.4955 43 | epoch: 40, mae: 26.10, mse: 28.73, rmae: 19.45%, rmse: 21.26%, r2: 0.6330 44 | epoch: 50, mae: 9.60, mse: 12.50, rmae: 7.22%, rmse: 9.45%, r2: 0.7562 45 | epoch: 60, mae: 59.74, mse: 63.42, rmae: 45.56%, rmse: 48.78%, r2: 0.3599 46 | epoch: 70, mae: 37.38, mse: 38.69, rmae: 28.28%, rmse: 29.39%, r2: 0.7938 47 | epoch: 80, mae: 20.19, mse: 23.86, rmae: 15.32%, rmse: 18.20%, r2: 0.6817 48 | epoch: 90, mae: 6.61, mse: 8.23, rmae: 5.03%, rmse: 6.27%, r2: 0.8099 49 | epoch: 100, mae: 15.40, mse: 16.61, rmae: 11.40%, rmse: 12.15%, r2: 0.8267 50 | epoch: 110, mae: 9.28, mse: 11.57, rmae: 7.10%, rmse: 8.94%, r2: 0.8604 51 | epoch: 120, mae: 32.80, mse: 35.75, rmae: 25.25%, rmse: 28.03%, r2: 0.5412 52 | epoch: 130, mae: 15.92, mse: 17.21, rmae: 11.82%, rmse: 12.60%, r2: 0.8691 53 | epoch: 140, mae: 13.56, mse: 15.18, rmae: 10.27%, rmse: 11.64%, r2: 0.8719 54 | epoch: 150, mae: 31.30, mse: 31.92, rmae: 23.49%, rmse: 23.94%, r2: 0.8889 55 | epoch: 160, mae: 17.83, mse: 19.52, rmae: 13.52%, rmse: 14.92%, r2: 0.8012 56 | epoch: 170, mae: 19.86, mse: 21.79, rmae: 14.93%, rmse: 16.46%, r2: 0.8380 57 | epoch: 180, mae: 13.61, mse: 14.98, rmae: 10.14%, rmse: 11.12%, r2: 0.8836 58 | epoch: 190, mae: 14.10, mse: 17.43, rmae: 10.85%, rmse: 13.43%, r2: 0.7943 59 | epoch: 200, mae: 6.50, mse: 8.42, rmae: 4.96%, rmse: 6.46%, r2: 0.8696 60 | epoch: 210, mae: 8.54, mse: 10.19, rmae: 6.45%, rmse: 7.80%, r2: 0.9013 61 | epoch: 220, mae: 6.20, mse: 7.58, rmae: 4.70%, rmse: 5.84%, r2: 0.9144 62 | epoch: 230, mae: 7.11, mse: 8.47, rmae: 5.29%, rmse: 6.31%, r2: 0.9136 63 | epoch: 240, mae: 5.78, mse: 7.04, rmae: 4.34%, rmse: 5.32%, r2: 0.9177 64 | epoch: 250, mae: 5.94, mse: 7.30, rmae: 4.53%, rmse: 5.65%, r2: 0.9102 65 | epoch: 260, mae: 8.57, mse: 9.79, rmae: 6.36%, rmse: 7.26%, r2: 0.9158 66 | epoch: 270, mae: 8.55, mse: 9.94, rmae: 6.44%, rmse: 7.58%, r2: 0.9175 67 | epoch: 280, mae: 11.03, mse: 12.29, rmae: 8.29%, rmse: 9.29%, r2: 0.9186 68 | epoch: 290, mae: 6.50, mse: 7.87, rmae: 4.89%, rmse: 5.95%, r2: 0.9150 69 | epoch: 300, mae: 4.59, mse: 5.66, rmae: 3.45%, rmse: 4.24%, r2: 0.9151 70 | epoch: 310, mae: 6.29, mse: 7.65, rmae: 4.69%, rmse: 5.74%, r2: 0.9214 71 | epoch: 320, mae: 6.30, mse: 7.69, rmae: 4.75%, rmse: 5.84%, r2: 0.9142 72 | epoch: 330, mae: 5.24, mse: 6.44, rmae: 3.96%, rmse: 4.91%, r2: 0.9132 73 | epoch: 340, mae: 4.80, mse: 5.70, rmae: 3.65%, rmse: 4.36%, r2: 0.9202 74 | epoch: 350, mae: 7.60, mse: 9.02, rmae: 5.72%, rmse: 6.89%, r2: 0.9073 75 | epoch: 360, mae: 5.61, mse: 6.87, rmae: 4.21%, rmse: 5.18%, r2: 0.9144 76 | epoch: 370, mae: 14.10, mse: 15.20, rmae: 10.62%, rmse: 11.49%, r2: 0.9105 77 | epoch: 380, mae: 5.44, mse: 6.64, rmae: 4.17%, rmse: 5.12%, r2: 0.9111 78 | epoch: 390, mae: 5.01, mse: 6.10, rmae: 3.80%, rmse: 4.63%, r2: 0.9105 79 | epoch: 400, mae: 15.86, mse: 16.83, rmae: 11.92%, rmse: 12.69%, r2: 0.9110 80 | epoch: 410, mae: 6.91, mse: 8.27, rmae: 5.19%, rmse: 6.27%, r2: 0.9165 81 | epoch: 420, mae: 6.47, mse: 7.77, rmae: 4.85%, rmse: 5.87%, r2: 0.9162 82 | epoch: 430, mae: 5.70, mse: 6.99, rmae: 4.27%, rmse: 5.28%, r2: 0.9155 83 | epoch: 440, mae: 5.49, mse: 6.74, rmae: 4.15%, rmse: 5.17%, r2: 0.9176 84 | epoch: 450, mae: 8.29, mse: 9.65, rmae: 6.20%, rmse: 7.26%, r2: 0.9143 85 | epoch: 460, mae: 6.48, mse: 7.81, rmae: 4.88%, rmse: 5.93%, r2: 0.9193 86 | epoch: 470, mae: 5.69, mse: 6.94, rmae: 4.28%, rmse: 5.27%, r2: 0.9179 87 | epoch: 480, mae: 7.49, mse: 8.81, rmae: 5.61%, rmse: 6.66%, r2: 0.9162 88 | epoch: 490, mae: 5.07, mse: 6.18, rmae: 3.83%, rmse: 4.71%, r2: 0.9177 89 | epoch: 500, mae: 7.57, mse: 8.95, rmae: 5.69%, rmse: 6.78%, r2: 0.9167 90 | best mae: 4.59, best mse: 5.66, best_rmae: 3.45, best_rmse: 4.24, best_r2: 0.9151 91 | overall best mae: 4.59, overall best mse: 5.66, overall best_rmae: 3.45, overall best_rmse: 4.24, overall best_r2: 0.9214 92 | -------------------------------------------------------------------------------- /snapshots/mtc/tasselnetv2plus_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500/tasselnetv2plus_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500.txt: -------------------------------------------------------------------------------- 1 | image_scale : 0.00392156862745098 2 | image_mean : [[[0.3859 0.4905 0.2895]]] 3 | image_std : [[[0.1718 0.1712 0.1518]]] 4 | scales : [0.7, 1, 1.3] 5 | shorter_side : 224 6 | data_dir : ./data/maize_counting_dataset 7 | dataset : mtc 8 | exp : tasselnetv2plus_normalinit_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500 9 | data_list : ./data/maize_counting_dataset/train.txt 10 | data_val_list : ./data/maize_counting_dataset/test.txt 11 | restore_from : ./snapshots/mtc/tasselnetv2plus_normalinit_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500/model_best.pth.tar 12 | snapshot_dir : ./snapshots/mtc/tasselnetv2plus_normalinit_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500 13 | result_dir : ./results/mtc/tasselnetv2plus_normalinit_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500 14 | save_output : False 15 | input_size : 64 16 | output_stride : 8 17 | resize_ratio : 0.125 18 | model : tasselnetv2plus 19 | width_mult : 1.0 20 | use_pretrained : False 21 | freeze_bn : False 22 | sync_bn : False 23 | use_nonlinear : False 24 | use_context : False 25 | use_squeeze : False 26 | optimizer : sgd 27 | batch_size : 9 28 | milestones : [200, 400] 29 | crop_size : (256, 256) 30 | evaluate_only : False 31 | learning_rate : 0.01 32 | momentum : 0.95 33 | weight_decay : 0.0005 34 | mult : 1 35 | num_epochs : 500 36 | num_workers : 0 37 | print_every : 10 38 | random_seed : 2020 39 | val_every : 10 40 | epoch: 10, mae: 8.63, mse: 14.44, relerr: 62.70%, relerr10: 22.07%, r2: 0.7150 41 | epoch: 20, mae: 11.16, mse: 15.78, relerr: 43.29%, relerr10: 35.18%, r2: 0.8175 42 | epoch: 30, mae: 14.76, mse: 21.61, relerr: 52.63%, relerr10: 44.74%, r2: 0.6139 43 | epoch: 40, mae: 8.72, mse: 14.23, relerr: 39.13%, relerr10: 23.12%, r2: 0.7990 44 | epoch: 50, mae: 16.41, mse: 23.71, relerr: 85.35%, relerr10: 39.88%, r2: 0.5240 45 | epoch: 60, mae: 17.25, mse: 19.73, relerr: 190.12%, relerr10: 51.44%, r2: 0.7861 46 | epoch: 70, mae: 8.46, mse: 12.55, relerr: 69.19%, relerr10: 23.82%, r2: 0.8175 47 | epoch: 80, mae: 11.28, mse: 17.45, relerr: 40.59%, relerr10: 38.37%, r2: 0.7293 48 | epoch: 90, mae: 13.38, mse: 18.58, relerr: 58.30%, relerr10: 36.34%, r2: 0.7935 49 | epoch: 100, mae: 11.40, mse: 15.84, relerr: 42.02%, relerr10: 36.31%, r2: 0.8323 50 | epoch: 110, mae: 11.83, mse: 17.81, relerr: 41.31%, relerr10: 39.48%, r2: 0.7446 51 | epoch: 120, mae: 8.48, mse: 11.62, relerr: 107.78%, relerr10: 26.57%, r2: 0.8379 52 | epoch: 130, mae: 6.12, mse: 9.62, relerr: 44.27%, relerr10: 17.42%, r2: 0.8823 53 | epoch: 140, mae: 6.45, mse: 9.86, relerr: 29.65%, relerr10: 20.40%, r2: 0.8844 54 | epoch: 150, mae: 7.68, mse: 12.73, relerr: 48.76%, relerr10: 20.25%, r2: 0.8309 55 | epoch: 160, mae: 6.04, mse: 9.11, relerr: 42.24%, relerr10: 18.50%, r2: 0.8841 56 | epoch: 170, mae: 6.22, mse: 8.95, relerr: 57.06%, relerr10: 17.26%, r2: 0.9062 57 | epoch: 180, mae: 8.02, mse: 14.07, relerr: 57.70%, relerr10: 19.45%, r2: 0.7666 58 | epoch: 190, mae: 7.54, mse: 11.11, relerr: 51.27%, relerr10: 22.86%, r2: 0.8334 59 | epoch: 200, mae: 9.34, mse: 14.16, relerr: 35.75%, relerr10: 27.77%, r2: 0.8446 60 | epoch: 210, mae: 5.61, mse: 10.14, relerr: 30.57%, relerr10: 17.31%, r2: 0.8704 61 | epoch: 220, mae: 6.35, mse: 10.65, relerr: 30.48%, relerr10: 19.50%, r2: 0.8771 62 | epoch: 230, mae: 5.38, mse: 9.51, relerr: 30.67%, relerr10: 16.03%, r2: 0.8841 63 | epoch: 240, mae: 5.71, mse: 10.06, relerr: 31.46%, relerr10: 16.90%, r2: 0.8765 64 | epoch: 250, mae: 5.55, mse: 10.05, relerr: 30.30%, relerr10: 15.84%, r2: 0.8767 65 | epoch: 260, mae: 5.89, mse: 10.63, relerr: 30.71%, relerr10: 16.56%, r2: 0.8707 66 | epoch: 270, mae: 5.75, mse: 10.01, relerr: 32.54%, relerr10: 16.37%, r2: 0.8785 67 | epoch: 280, mae: 5.77, mse: 9.94, relerr: 29.51%, relerr10: 16.51%, r2: 0.8939 68 | epoch: 290, mae: 5.30, mse: 9.50, relerr: 31.07%, relerr10: 14.57%, r2: 0.8900 69 | epoch: 300, mae: 6.27, mse: 10.51, relerr: 34.51%, relerr10: 17.63%, r2: 0.8783 70 | epoch: 310, mae: 5.09, mse: 9.06, relerr: 33.81%, relerr10: 14.09%, r2: 0.8880 71 | epoch: 320, mae: 5.16, mse: 9.09, relerr: 33.77%, relerr10: 14.54%, r2: 0.8926 72 | epoch: 330, mae: 5.93, mse: 10.70, relerr: 30.02%, relerr10: 17.17%, r2: 0.8641 73 | epoch: 340, mae: 5.37, mse: 9.95, relerr: 28.80%, relerr10: 15.50%, r2: 0.8748 74 | epoch: 350, mae: 5.87, mse: 10.64, relerr: 33.15%, relerr10: 16.06%, r2: 0.8594 75 | epoch: 360, mae: 5.81, mse: 9.96, relerr: 30.95%, relerr10: 15.51%, r2: 0.8951 76 | epoch: 370, mae: 6.62, mse: 11.25, relerr: 35.86%, relerr10: 17.99%, r2: 0.8696 77 | epoch: 380, mae: 5.68, mse: 10.50, relerr: 32.99%, relerr10: 15.58%, r2: 0.8603 78 | epoch: 390, mae: 5.34, mse: 9.74, relerr: 31.56%, relerr10: 14.74%, r2: 0.8725 79 | epoch: 400, mae: 5.46, mse: 9.66, relerr: 29.28%, relerr10: 15.28%, r2: 0.8939 80 | epoch: 410, mae: 5.89, mse: 10.64, relerr: 28.17%, relerr10: 17.17%, r2: 0.8688 81 | epoch: 420, mae: 5.55, mse: 10.15, relerr: 30.31%, relerr10: 15.38%, r2: 0.8741 82 | epoch: 430, mae: 5.50, mse: 10.24, relerr: 30.20%, relerr10: 15.26%, r2: 0.8677 83 | epoch: 440, mae: 5.53, mse: 10.13, relerr: 31.68%, relerr10: 14.97%, r2: 0.8751 84 | epoch: 450, mae: 5.56, mse: 10.36, relerr: 30.76%, relerr10: 14.98%, r2: 0.8693 85 | epoch: 460, mae: 5.44, mse: 9.88, relerr: 31.79%, relerr10: 14.66%, r2: 0.8814 86 | epoch: 470, mae: 5.50, mse: 10.03, relerr: 32.37%, relerr10: 14.67%, r2: 0.8778 87 | epoch: 480, mae: 5.52, mse: 10.09, relerr: 33.53%, relerr10: 14.71%, r2: 0.8753 88 | epoch: 490, mae: 5.96, mse: 10.62, relerr: 30.87%, relerr10: 16.10%, r2: 0.8741 89 | epoch: 500, mae: 5.58, mse: 10.22, relerr: 29.42%, relerr10: 15.37%, r2: 0.8765 90 | best mae: 5.09, best mse: 9.06, best_relerr: 33.81, best_relerr10: 14.09, best_r2: 0.8880 91 | overall best mae: 5.09, overall best mse: 8.95, overall best_relerr: 28.17, overall best_relerr10: 14.09, overall best_r2: 0.9062 92 | -------------------------------------------------------------------------------- /snapshots/shc/tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500.txt: -------------------------------------------------------------------------------- 1 | image_scale : 0.00392156862745098 2 | image_mean : [[[0.3714 0.3609 0.2386]]] 3 | image_std : [[[0.2705 0.2567 0.2161]]] 4 | scales : [0.7, 1, 1.3] 5 | shorter_side : 224 6 | data_dir : ./data/sorghum_head_counting_dataset 7 | dataset : shc 8 | exp : tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500 9 | data_list : ./data/sorghum_head_counting_dataset/dataset2_train.txt 10 | data_val_list : ./data/sorghum_head_counting_dataset/dataset2_test.txt 11 | restore_from : ./snapshots/shc/tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500/model_best.pth.tar 12 | snapshot_dir : ./snapshots/shc/tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500 13 | result_dir : ./results/shc/tasselnetv2plus_dataset2_rf110_i64o8_r1_crop1024_lr-2_bs10_epoch500 14 | save_output : False 15 | input_size : 64 16 | output_stride : 8 17 | resize_ratio : 1.0 18 | model : tasselnetv2plus 19 | width_mult : 1.0 20 | use_pretrained : False 21 | freeze_bn : False 22 | sync_bn : False 23 | use_nonlinear : False 24 | use_context : False 25 | use_squeeze : False 26 | optimizer : sgd 27 | batch_size : 5 28 | milestones : [200, 400] 29 | crop_size : (256, 1024) 30 | evaluate_only : False 31 | learning_rate : 0.01 32 | momentum : 0.95 33 | weight_decay : 0.0005 34 | mult : 1 35 | num_epochs : 500 36 | num_workers : 0 37 | print_every : 4 38 | random_seed : 2020 39 | val_every : 10 40 | epoch: 10, mae: 41.29, mse: 46.29, relerr: 41.09%, relerr10: 41.09%, r2: 0.0691 41 | epoch: 20, mae: 11.44, mse: 12.30, relerr: 11.18%, relerr10: 11.18%, r2: 0.4129 42 | epoch: 30, mae: 39.58, mse: 39.99, relerr: 39.08%, relerr10: 39.08%, r2: 0.3798 43 | epoch: 40, mae: 22.82, mse: 23.76, relerr: 22.41%, relerr10: 22.41%, r2: 0.2649 44 | epoch: 50, mae: 43.68, mse: 44.28, relerr: 43.37%, relerr10: 43.37%, r2: 0.5122 45 | epoch: 60, mae: 33.05, mse: 34.23, relerr: 32.60%, relerr10: 32.60%, r2: 0.1628 46 | epoch: 70, mae: 24.90, mse: 27.40, relerr: 24.67%, relerr10: 24.67%, r2: 0.3279 47 | epoch: 80, mae: 45.99, mse: 47.63, relerr: 45.53%, relerr10: 45.53%, r2: 0.3714 48 | epoch: 90, mae: 23.05, mse: 28.55, relerr: 22.89%, relerr10: 22.89%, r2: 0.1397 49 | epoch: 100, mae: 13.39, mse: 17.55, relerr: 13.28%, relerr10: 13.28%, r2: 0.3006 50 | epoch: 110, mae: 15.50, mse: 19.12, relerr: 15.33%, relerr10: 15.33%, r2: 0.3328 51 | epoch: 120, mae: 3.58, mse: 4.78, relerr: 3.57%, relerr10: 3.57%, r2: 0.6767 52 | epoch: 130, mae: 10.46, mse: 12.84, relerr: 10.39%, relerr10: 10.39%, r2: 0.4653 53 | epoch: 140, mae: 14.22, mse: 14.92, relerr: 14.08%, relerr10: 14.08%, r2: 0.3705 54 | epoch: 150, mae: 13.82, mse: 14.90, relerr: 13.69%, relerr10: 13.69%, r2: 0.6517 55 | epoch: 160, mae: 17.76, mse: 18.84, relerr: 17.64%, relerr10: 17.64%, r2: 0.5625 56 | epoch: 170, mae: 11.72, mse: 13.31, relerr: 11.70%, relerr10: 11.70%, r2: 0.4438 57 | epoch: 180, mae: 7.93, mse: 8.79, relerr: 7.86%, relerr10: 7.86%, r2: 0.6470 58 | epoch: 190, mae: 49.86, mse: 52.81, relerr: 49.36%, relerr10: 49.36%, r2: 0.2279 59 | epoch: 200, mae: 7.45, mse: 12.08, relerr: 7.44%, relerr10: 7.44%, r2: 0.3438 60 | epoch: 210, mae: 4.44, mse: 7.03, relerr: 4.44%, relerr10: 4.44%, r2: 0.5064 61 | epoch: 220, mae: 5.59, mse: 8.52, relerr: 5.56%, relerr10: 5.56%, r2: 0.4720 62 | epoch: 230, mae: 7.24, mse: 10.56, relerr: 7.21%, relerr10: 7.21%, r2: 0.4450 63 | epoch: 240, mae: 5.67, mse: 8.64, relerr: 5.64%, relerr10: 5.64%, r2: 0.4703 64 | epoch: 250, mae: 9.52, mse: 12.68, relerr: 9.44%, relerr10: 9.44%, r2: 0.4255 65 | epoch: 260, mae: 10.70, mse: 13.29, relerr: 10.61%, relerr10: 10.61%, r2: 0.4551 66 | epoch: 270, mae: 11.42, mse: 13.68, relerr: 11.31%, relerr10: 11.31%, r2: 0.4940 67 | epoch: 280, mae: 7.11, mse: 9.60, relerr: 7.06%, relerr10: 7.06%, r2: 0.5355 68 | epoch: 290, mae: 8.06, mse: 11.13, relerr: 7.99%, relerr10: 7.99%, r2: 0.4615 69 | epoch: 300, mae: 5.74, mse: 8.46, relerr: 5.68%, relerr10: 5.68%, r2: 0.5134 70 | epoch: 310, mae: 5.50, mse: 8.85, relerr: 5.49%, relerr10: 5.49%, r2: 0.4416 71 | epoch: 320, mae: 10.52, mse: 13.70, relerr: 10.42%, relerr10: 10.42%, r2: 0.4194 72 | epoch: 330, mae: 5.66, mse: 8.59, relerr: 5.60%, relerr10: 5.60%, r2: 0.4850 73 | epoch: 340, mae: 5.21, mse: 7.46, relerr: 5.18%, relerr10: 5.18%, r2: 0.5637 74 | epoch: 350, mae: 4.21, mse: 6.45, relerr: 4.18%, relerr10: 4.18%, r2: 0.5591 75 | epoch: 360, mae: 5.18, mse: 8.13, relerr: 5.18%, relerr10: 5.18%, r2: 0.4859 76 | epoch: 370, mae: 8.50, mse: 11.80, relerr: 8.42%, relerr10: 8.42%, r2: 0.4413 77 | epoch: 380, mae: 14.24, mse: 16.45, relerr: 14.08%, relerr10: 14.08%, r2: 0.4872 78 | epoch: 390, mae: 4.61, mse: 7.21, relerr: 4.61%, relerr10: 4.61%, r2: 0.5192 79 | epoch: 400, mae: 8.65, mse: 11.75, relerr: 8.57%, relerr10: 8.57%, r2: 0.4575 80 | epoch: 410, mae: 7.96, mse: 10.68, relerr: 7.89%, relerr10: 7.89%, r2: 0.5154 81 | epoch: 420, mae: 6.87, mse: 9.60, relerr: 6.82%, relerr10: 6.82%, r2: 0.5159 82 | epoch: 430, mae: 6.66, mse: 9.32, relerr: 6.60%, relerr10: 6.60%, r2: 0.5228 83 | epoch: 440, mae: 8.05, mse: 10.71, relerr: 7.97%, relerr10: 7.97%, r2: 0.5148 84 | epoch: 450, mae: 6.86, mse: 9.48, relerr: 6.81%, relerr10: 6.81%, r2: 0.5299 85 | epoch: 460, mae: 6.90, mse: 9.78, relerr: 6.86%, relerr10: 6.86%, r2: 0.4983 86 | epoch: 470, mae: 9.23, mse: 11.75, relerr: 9.14%, relerr10: 9.14%, r2: 0.5070 87 | epoch: 480, mae: 6.56, mse: 9.08, relerr: 6.52%, relerr10: 6.52%, r2: 0.5389 88 | epoch: 490, mae: 8.32, mse: 10.90, relerr: 8.25%, relerr10: 8.25%, r2: 0.5152 89 | epoch: 500, mae: 8.80, mse: 11.62, relerr: 8.71%, relerr10: 8.71%, r2: 0.4849 90 | best mae: 3.58, best mse: 4.78, best_relerr: 3.57, best_relerr10: 3.57, best_r2: 0.6767 91 | overall best mae: 3.58, overall best mse: 4.78, overall best_relerr: 3.57, overall best_relerr10: 3.57, overall best_r2: 0.6767 92 | epoch: 120, mae: 3.58, mse: 4.78, relerr: 3.57%, relerr10: 3.57%, r2: 0.6767 93 | -------------------------------------------------------------------------------- /hlnet.py: -------------------------------------------------------------------------------- 1 | # -*- coding: utf-8 -*- 2 | """ 3 | @author: hao lu 4 | """ 5 | 6 | import torch 7 | import torch.nn as nn 8 | import torch.nn.functional as F 9 | from utils import * 10 | 11 | 12 | class Encoder(nn.Module): 13 | def __init__(self, arc='tasselnetv2plus'): 14 | super(Encoder, self).__init__() 15 | if arc == 'tasselnetv2': 16 | self.encoder = nn.Sequential( 17 | nn.Conv2d(3, 16, 3, padding=1, bias=False), 18 | nn.BatchNorm2d(16), 19 | nn.ReLU(inplace=True), 20 | nn.MaxPool2d((2, 2), stride=2), 21 | nn.Conv2d(16, 32, 3, padding=1, bias=False), 22 | nn.BatchNorm2d(32), 23 | nn.ReLU(inplace=True), 24 | nn.MaxPool2d((2, 2), stride=2), 25 | nn.Conv2d(32, 64, 3, padding=1, bias=False), 26 | nn.BatchNorm2d(64), 27 | nn.ReLU(inplace=True), 28 | nn.MaxPool2d((2, 2), stride=2), 29 | nn.Conv2d(64, 128, 3, padding=1, bias=False), 30 | nn.BatchNorm2d(128), 31 | nn.ReLU(inplace=True), 32 | nn.Conv2d(128, 128, 3, padding=1, bias=False), 33 | nn.BatchNorm2d(128), 34 | nn.ReLU(inplace=True), 35 | ) 36 | elif arc == 'tasselnetv2plus': 37 | self.encoder = nn.Sequential( 38 | nn.Conv2d(3, 16, 3, padding=1, bias=False), 39 | nn.BatchNorm2d(16), 40 | nn.ReLU(inplace=True), 41 | nn.MaxPool2d((2, 2), stride=2), 42 | nn.Conv2d(16, 32, 3, padding=1, bias=False), 43 | nn.BatchNorm2d(32), 44 | nn.ReLU(inplace=True), 45 | nn.MaxPool2d((2, 2), stride=2), 46 | nn.Conv2d(32, 64, 3, padding=1, bias=False), 47 | nn.BatchNorm2d(64), 48 | nn.ReLU(inplace=True), 49 | nn.MaxPool2d((2, 2), stride=2), 50 | nn.Conv2d(64, 128, 3, padding=1, bias=False), 51 | nn.BatchNorm2d(128), 52 | nn.ReLU(inplace=True), 53 | nn.Conv2d(128, 128, 3, padding=1, bias=False), 54 | nn.BatchNorm2d(128), 55 | nn.ReLU(inplace=True) 56 | ) 57 | else: 58 | raise NotImplementedError 59 | 60 | def forward(self, x): 61 | x = self.encoder(x) 62 | return x 63 | 64 | 65 | class Counter(nn.Module): 66 | def __init__(self, arc='tasselnetv2plus', input_size=64, output_stride=8): 67 | super(Counter, self).__init__() 68 | k = int(input_size / 8) 69 | avg_pool_stride = int(output_stride / 8) 70 | 71 | if arc == 'tasselnetv2': 72 | self.counter = nn.Sequential( 73 | nn.Conv2d(128, 128, (k, k), bias=False), 74 | nn.BatchNorm2d(128), 75 | nn.ReLU(inplace=True), 76 | nn.Conv2d(128, 128, 1, bias=False), 77 | nn.BatchNorm2d(128), 78 | nn.ReLU(inplace=True), 79 | nn.Conv2d(128, 1, 1) 80 | ) 81 | elif arc == 'tasselnetv2plus': 82 | self.counter = nn.Sequential( 83 | nn.AvgPool2d((k, k), stride=avg_pool_stride), 84 | nn.Conv2d(128, 128, 1, bias=False), 85 | nn.BatchNorm2d(128), 86 | nn.ReLU(inplace=True), 87 | nn.Conv2d(128, 1, 1) 88 | ) 89 | else: 90 | raise NotImplementedError 91 | 92 | def forward(self, x): 93 | x = self.counter(x) 94 | return x 95 | 96 | 97 | class Normalizer: 98 | @staticmethod 99 | def cpu_normalizer(x, imh, imw, insz, os): 100 | # CPU normalization 101 | bs = x.size()[0] 102 | normx = np.zeros((imh, imw)) 103 | norm_vec = dense_sample2d(normx, insz, os).astype(np.float32) 104 | x = x.cpu().detach().numpy().reshape(bs, -1) * norm_vec 105 | return x 106 | 107 | @staticmethod 108 | def gpu_normalizer(x, imh, imw, insz, os): 109 | _, _, h, w = x.size() 110 | accm = torch.cuda.FloatTensor(1, insz*insz, h*w).fill_(1) 111 | accm = F.fold(accm, (imh, imw), kernel_size=insz, stride=os) 112 | accm = 1 / accm 113 | accm /= insz**2 114 | accm = F.unfold(accm, kernel_size=insz, stride=os).sum(1).view(1, 1, h, w) 115 | x *= accm 116 | return x.squeeze().cpu().detach().numpy() 117 | 118 | 119 | class CountingModels(nn.Module): 120 | def __init__(self, arc='tasselnetv2plus', input_size=64, output_stride=8): 121 | super(CountingModels, self).__init__() 122 | self.input_size = input_size 123 | self.output_stride = output_stride 124 | 125 | self.encoder = Encoder(arc) 126 | self.counter = Counter(arc, input_size, output_stride) 127 | if arc == 'tasselnetv2': 128 | self.normalizer = Normalizer.cpu_normalizer 129 | elif arc == 'tasselnetv2plus': 130 | self.normalizer = Normalizer.gpu_normalizer 131 | 132 | self.weight_init() 133 | 134 | def forward(self, x, is_normalize=True): 135 | imh, imw = x.size()[2:] 136 | x = self.encoder(x) 137 | x = self.counter(x) 138 | if is_normalize: 139 | x = self.normalizer(x, imh, imw, self.input_size, self.output_stride) 140 | return x 141 | 142 | def weight_init(self): 143 | for m in self.modules(): 144 | if isinstance(m, nn.Conv2d): 145 | nn.init.normal_(m.weight, std=0.01) 146 | # nn.init.kaiming_uniform_( 147 | # m.weight, 148 | # mode='fan_in', 149 | # nonlinearity='relu' 150 | # ) 151 | if m.bias is not None: 152 | nn.init.constant_(m.bias, 0) 153 | elif isinstance(m, nn.BatchNorm2d): 154 | nn.init.constant_(m.weight, 1) 155 | nn.init.constant_(m.bias, 0) 156 | 157 | 158 | if __name__ == "__main__": 159 | 160 | from time import time 161 | 162 | insz, os = 64, 8 163 | imH, imW = 1080, 1920 164 | net = CountingModels(arc='tasselnetv2plus', input_size=insz, output_stride=os).cuda() 165 | with torch.no_grad(): 166 | net.eval() 167 | x = torch.randn(1, 3, imH, imW).cuda() 168 | y = net(x) 169 | print(y.shape) 170 | 171 | import numpy as np 172 | 173 | with torch.no_grad(): 174 | frame_rate = np.zeros((100, 1)) 175 | 176 | for i in range(100): 177 | x = torch.randn(1, 3, imH, imW).cuda() 178 | torch.cuda.synchronize() 179 | start = time() 180 | 181 | y = net(x) 182 | 183 | torch.cuda.synchronize() 184 | end = time() 185 | 186 | running_frame_rate = 1 * float(1 / (end - start)) 187 | frame_rate[i] = running_frame_rate 188 | print(np.mean(frame_rate)) -------------------------------------------------------------------------------- /data/wheat_ears_counting_dataset/train.txt: -------------------------------------------------------------------------------- 1 | /train/images/2091.JPG /train/labels/2091.xml 2 | /train/images/2163.JPG /train/labels/2163.xml 3 | /train/images/2181.JPG /train/labels/2181.xml 4 | /train/images/5102.JPG /train/labels/5102.xml 5 | /train/images/2033.JPG /train/labels/2033.xml 6 | /train/images/6081.JPG /train/labels/6081.xml 7 | /train/images/5111.JPG /train/labels/5111.xml 8 | /train/images/6101.JPG /train/labels/6101.xml 9 | /train/images/4153.JPG /train/labels/4153.xml 10 | /train/images/2151.JPG /train/labels/2151.xml 11 | /train/images/6031.JPG /train/labels/6031.xml 12 | /train/images/1142.JPG /train/labels/1142.xml 13 | /train/images/2192.JPG /train/labels/2192.xml 14 | /train/images/2092.JPG /train/labels/2092.xml 15 | /train/images/6032.JPG /train/labels/6032.xml 16 | /train/images/3012.JPG /train/labels/3012.xml 17 | /train/images/5151.JPG /train/labels/5151.xml 18 | /train/images/2012.JPG /train/labels/2012.xml 19 | /train/images/6082.JPG /train/labels/6082.xml 20 | /train/images/6191.JPG /train/labels/6191.xml 21 | /train/images/5011.JPG /train/labels/5011.xml 22 | /train/images/5112.JPG /train/labels/5112.xml 23 | /train/images/3081.JPG /train/labels/3081.xml 24 | /train/images/6121.JPG /train/labels/6121.xml 25 | /train/images/6151.JPG /train/labels/6151.xml 26 | /train/images/4182.JPG /train/labels/4182.xml 27 | /train/images/4071.JPG /train/labels/4071.xml 28 | /train/images/4142.JPG /train/labels/4142.xml 29 | /train/images/5022.JPG /train/labels/5022.xml 30 | /train/images/1191.JPG /train/labels/1191.xml 31 | /train/images/2132.JPG /train/labels/2132.xml 32 | /train/images/5152.JPG /train/labels/5152.xml 33 | /train/images/4141.JPG /train/labels/4141.xml 34 | /train/images/3011.JPG /train/labels/3011.xml 35 | /train/images/5032.JPG /train/labels/5032.xml 36 | /train/images/6122.JPG /train/labels/6122.xml 37 | /train/images/4032.JPG /train/labels/4032.xml 38 | /train/images/3122.JPG /train/labels/3122.xml 39 | /train/images/4181.JPG /train/labels/4181.xml 40 | /train/images/1082.JPG /train/labels/1082.xml 41 | /train/images/5132.JPG /train/labels/5132.xml 42 | /train/images/2062.JPG /train/labels/2062.xml 43 | /train/images/5161.JPG /train/labels/5161.xml 44 | /train/images/3032.JPG /train/labels/3032.xml 45 | /train/images/2041.JPG /train/labels/2041.xml 46 | /train/images/5201.JPG /train/labels/5201.xml 47 | /train/images/1073.JPG /train/labels/1073.xml 48 | /train/images/1182.JPG /train/labels/1182.xml 49 | /train/images/4092.JPG /train/labels/4092.xml 50 | /train/images/6171.JPG /train/labels/6171.xml 51 | /train/images/1021.JPG /train/labels/1021.xml 52 | /train/images/2131.JPG /train/labels/2131.xml 53 | /train/images/3061.JPG /train/labels/3061.xml 54 | /train/images/3111.JPG /train/labels/3111.xml 55 | /train/images/3052.JPG /train/labels/3052.xml 56 | /train/images/1141.JPG /train/labels/1141.xml 57 | /train/images/5131.JPG /train/labels/5131.xml 58 | /train/images/6202.JPG /train/labels/6202.xml 59 | /train/images/1192.JPG /train/labels/1192.xml 60 | /train/images/5031.JPG /train/labels/5031.xml 61 | /train/images/1042.JPG /train/labels/1042.xml 62 | /train/images/6072.JPG /train/labels/6072.xml 63 | /train/images/6141.JPG /train/labels/6141.xml 64 | /train/images/2171.JPG /train/labels/2171.xml 65 | /train/images/2191.JPG /train/labels/2191.xml 66 | /train/images/3051.JPG /train/labels/3051.xml 67 | /train/images/3121.JPG /train/labels/3121.xml 68 | /train/images/2082.JPG /train/labels/2082.xml 69 | /train/images/5071.JPG /train/labels/5071.xml 70 | /train/images/3181.JPG /train/labels/3181.xml 71 | /train/images/4201.JPG /train/labels/4201.xml 72 | /train/images/3082.JPG /train/labels/3082.xml 73 | /train/images/4161.JPG /train/labels/4161.xml 74 | /train/images/5101.JPG /train/labels/5101.xml 75 | /train/images/6102.JPG /train/labels/6102.xml 76 | /train/images/6181.JPG /train/labels/6181.xml 77 | /train/images/1132.JPG /train/labels/1132.xml 78 | /train/images/5041.JPG /train/labels/5041.xml 79 | /train/images/4111.JPG /train/labels/4111.xml 80 | /train/images/2172.JPG /train/labels/2172.xml 81 | /train/images/4012.JPG /train/labels/4012.xml 82 | /train/images/1152.JPG /train/labels/1152.xml 83 | /train/images/5012.JPG /train/labels/5012.xml 84 | /train/images/3092.JPG /train/labels/3092.xml 85 | /train/images/2081.JPG /train/labels/2081.xml 86 | /train/images/2061.JPG /train/labels/2061.xml 87 | /train/images/1032.JPG /train/labels/1032.xml 88 | /train/images/4202.JPG /train/labels/4202.xml 89 | /train/images/1031.JPG /train/labels/1031.xml 90 | /train/images/1091.JPG /train/labels/1091.xml 91 | /train/images/3182.JPG /train/labels/3182.xml 92 | /train/images/3042.JPG /train/labels/3042.xml 93 | /train/images/1202.JPG /train/labels/1202.xml 94 | /train/images/3041.JPG /train/labels/3041.xml 95 | /train/images/6182.JPG /train/labels/6182.xml 96 | /train/images/4051.JPG /train/labels/4051.xml 97 | /train/images/4162.JPG /train/labels/4162.xml 98 | /train/images/6041.JPG /train/labels/6041.xml 99 | /train/images/1022.JPG /train/labels/1022.xml 100 | /train/images/3021.JPG /train/labels/3021.xml 101 | /train/images/5042.JPG /train/labels/5042.xml 102 | /train/images/6142.JPG /train/labels/6142.xml 103 | /train/images/2141.JPG /train/labels/2141.xml 104 | /train/images/1151.JPG /train/labels/1151.xml 105 | /train/images/5162.JPG /train/labels/5162.xml 106 | /train/images/2142.JPG /train/labels/2142.xml 107 | /train/images/2182.JPG /train/labels/2182.xml 108 | /train/images/1081.JPG /train/labels/1081.xml 109 | /train/images/1092.JPG /train/labels/1092.xml 110 | /train/images/4072.JPG /train/labels/4072.xml 111 | /train/images/4042.JPG /train/labels/4042.xml 112 | /train/images/1111.JPG /train/labels/1111.xml 113 | /train/images/2122.JPG /train/labels/2122.xml 114 | /train/images/4101.JPG /train/labels/4101.xml 115 | /train/images/3132.JPG /train/labels/3132.xml 116 | /train/images/4152.JPG /train/labels/4152.xml 117 | /train/images/4011.JPG /train/labels/4011.xml 118 | /train/images/4052.JPG /train/labels/4052.xml 119 | /train/images/6172.JPG /train/labels/6172.xml 120 | /train/images/5021.JPG /train/labels/5021.xml 121 | /train/images/4102.JPG /train/labels/4102.xml 122 | /train/images/6092.JPG /train/labels/6092.xml 123 | /train/images/4031.JPG /train/labels/4031.xml 124 | /train/images/6071.JPG /train/labels/6071.xml 125 | /train/images/3141.JPG /train/labels/3141.xml 126 | /train/images/6192.JPG /train/labels/6192.xml 127 | /train/images/4151.JPG /train/labels/4151.xml 128 | /train/images/6152.JPG /train/labels/6152.xml 129 | /train/images/3031.JPG /train/labels/3031.xml 130 | /train/images/1112.JPG /train/labels/1112.xml 131 | /train/images/6111.JPG /train/labels/6111.xml 132 | /train/images/3133.JPG /train/labels/3133.xml 133 | /train/images/3062.JPG /train/labels/3062.xml 134 | /train/images/6112.JPG /train/labels/6112.xml 135 | /train/images/5082.JPG /train/labels/5082.xml 136 | /train/images/1071.JPG /train/labels/1071.xml 137 | /train/images/5202.JPG /train/labels/5202.xml 138 | /train/images/6091.JPG /train/labels/6091.xml 139 | /train/images/3112.JPG /train/labels/3112.xml 140 | /train/images/1181.JPG /train/labels/1181.xml 141 | /train/images/1171.JPG /train/labels/1171.xml 142 | /train/images/5052.JPG /train/labels/5052.xml 143 | /train/images/3022.JPG /train/labels/3022.xml 144 | /train/images/6042.JPG /train/labels/6042.xml 145 | /train/images/4062.JPG /train/labels/4062.xml 146 | /train/images/2011.JPG /train/labels/2011.xml 147 | /train/images/5072.JPG /train/labels/5072.xml 148 | /train/images/5081.JPG /train/labels/5081.xml 149 | /train/images/4091.JPG /train/labels/4091.xml 150 | /train/images/3172.JPG /train/labels/3172.xml 151 | /train/images/6201.JPG /train/labels/6201.xml 152 | /train/images/5192.JPG /train/labels/5192.xml 153 | /train/images/5191.JPG /train/labels/5191.xml 154 | /train/images/1131.JPG /train/labels/1131.xml 155 | /train/images/3142.JPG /train/labels/3142.xml 156 | /train/images/1172.JPG /train/labels/1172.xml 157 | /train/images/3091.JPG /train/labels/3091.xml 158 | /train/images/2152.JPG /train/labels/2152.xml 159 | /train/images/3171.JPG /train/labels/3171.xml 160 | /train/images/2121.JPG /train/labels/2121.xml 161 | /train/images/4061.JPG /train/labels/4061.xml 162 | /train/images/4112.JPG /train/labels/4112.xml 163 | /train/images/1041.JPG /train/labels/1041.xml 164 | /train/images/1201.JPG /train/labels/1201.xml 165 | /train/images/4041.JPG /train/labels/4041.xml -------------------------------------------------------------------------------- /data/maize_counting_dataset/test.txt: -------------------------------------------------------------------------------- 1 | /test/images/T0001_XM_20110807130247_01.jpg /test/labels/T0001_XM_20110807130247_01.mat 2 | /test/images/T0001_XM_20110813160244_01.jpg /test/labels/T0001_XM_20110813160244_01.mat 3 | /test/images/T0006_XM_20140811160049_01.jpg /test/labels/T0006_XM_20140811160049_01.mat 4 | /test/images/T0001_XM_20120802140257_02.jpg /test/labels/T0001_XM_20120802140257_02.mat 5 | /test/images/T0001_XM_20120801100256_02.jpg /test/labels/T0001_XM_20120801100256_02.mat 6 | /test/images/T0001_XM_20110812100243_01.jpg /test/labels/T0001_XM_20110812100243_01.mat 7 | /test/images/XAM02_YM_20150727100010_01.jpg /test/labels/XAM02_YM_20150727100010_01.mat 8 | /test/images/T0002_XM_20110728110224_01.jpg /test/labels/T0002_XM_20110728110224_01.mat 9 | /test/images/T0006_YM_20140818180046_01.jpg /test/labels/T0006_YM_20140818180046_01.mat 10 | /test/images/T0006_YM_20140818170051_01.jpg /test/labels/T0006_YM_20140818170051_01.mat 11 | /test/images/T0006_YM_20140818090047_01.jpg /test/labels/T0006_YM_20140818090047_01.mat 12 | /test/images/T0006_YM_20140816140055_01.jpg /test/labels/T0006_YM_20140816140055_01.mat 13 | /test/images/T0002_XM_20110802100228_01.jpg /test/labels/T0002_XM_20110802100228_01.mat 14 | /test/images/XAM02_YM_20150725100005_01.jpg /test/labels/XAM02_YM_20150725100005_01.mat 15 | /test/images/T0001_YM_20100803160241_01.jpg /test/labels/T0001_YM_20100803160241_01.mat 16 | /test/images/T0001_YM_20100804100235_01.jpg /test/labels/T0001_YM_20100804100235_01.mat 17 | /test/images/T0001_YM_20100804140239_01.jpg /test/labels/T0001_YM_20100804140239_01.mat 18 | /test/images/XAM05_YM_20150731160225_01.jpg /test/labels/XAM05_YM_20150731160225_01.mat 19 | /test/images/T0002_XM_20110727100223_01.jpg /test/labels/T0002_XM_20110727100223_01.mat 20 | /test/images/T0001_XM_20120805090254_02.jpg /test/labels/T0001_XM_20120805090254_02.mat 21 | /test/images/T0001_XM_20110811160245_01.jpg /test/labels/T0001_XM_20110811160245_01.mat 22 | /test/images/T0006_XM_20140809120051_01.jpg /test/labels/T0006_XM_20140809120051_01.mat 23 | /test/images/XAM05_YM_20150726160228_01.jpg /test/labels/XAM05_YM_20150726160228_01.mat 24 | /test/images/T0001_XM_20120806150256_02.jpg /test/labels/T0001_XM_20120806150256_02.mat 25 | /test/images/T0002_XM_20110802090224_01.jpg /test/labels/T0002_XM_20110802090224_01.mat 26 | /test/images/T0001_XM_20130810100251_02.jpg /test/labels/T0001_XM_20130810100251_02.mat 27 | /test/images/T0001_XM_20120802110255_02.jpg /test/labels/T0001_XM_20120802110255_02.mat 28 | /test/images/T0001_XM_20110809120248_01.jpg /test/labels/T0001_XM_20110809120248_01.mat 29 | /test/images/T0002_XM_20110802120223_01.jpg /test/labels/T0002_XM_20110802120223_01.mat 30 | /test/images/XAM05_YM_20150801160224_01.jpg /test/labels/XAM05_YM_20150801160224_01.mat 31 | /test/images/T0006_XM_20140811170049_01.jpg /test/labels/T0006_XM_20140811170049_01.mat 32 | /test/images/T0006_XM_20140808090048_01.jpg /test/labels/T0006_XM_20140808090048_01.mat 33 | /test/images/T0001_XM_20120801160256_02.jpg /test/labels/T0001_XM_20120801160256_02.mat 34 | /test/images/T0001_XM_20120805150254_02.jpg /test/labels/T0001_XM_20120805150254_02.mat 35 | /test/images/T0001_XM_20120807090254_02.jpg /test/labels/T0001_XM_20120807090254_02.mat 36 | /test/images/T0001_XM_20110808160246_01.jpg /test/labels/T0001_XM_20110808160246_01.mat 37 | /test/images/T0001_XM_20110808100247_01.jpg /test/labels/T0001_XM_20110808100247_01.mat 38 | /test/images/T0006_XM_20140812080052_01.jpg /test/labels/T0006_XM_20140812080052_01.mat 39 | /test/images/T0006_YM_20140816120051_01.jpg /test/labels/T0006_YM_20140816120051_01.mat 40 | /test/images/T0006_XM_20140813110051_01.jpg /test/labels/T0006_XM_20140813110051_01.mat 41 | /test/images/XAM02_YM_20150731100011_01.jpg /test/labels/XAM02_YM_20150731100011_01.mat 42 | /test/images/T0006_YM_20140817100046_01.jpg /test/labels/T0006_YM_20140817100046_01.mat 43 | /test/images/T0006_YM_20140818110052_01.jpg /test/labels/T0006_YM_20140818110052_01.mat 44 | /test/images/T0002_XM_20110728130225_01.jpg /test/labels/T0002_XM_20110728130225_01.mat 45 | /test/images/T0002_XM_20110729160224_01.jpg /test/labels/T0002_XM_20110729160224_01.mat 46 | /test/images/T0006_XM_20140814130054_01.jpg /test/labels/T0006_XM_20140814130054_01.mat 47 | /test/images/T0006_XM_20140813160048_01.jpg /test/labels/T0006_XM_20140813160048_01.mat 48 | /test/images/XAM05_YM_20150725160227_01.jpg /test/labels/XAM05_YM_20150725160227_01.mat 49 | /test/images/T0002_XM_20110728090224_01.jpg /test/labels/T0002_XM_20110728090224_01.mat 50 | /test/images/T0001_YM_20100808110235_01.jpg /test/labels/T0001_YM_20100808110235_01.mat 51 | /test/images/T0001_XM_20120806130253_02.jpg /test/labels/T0001_XM_20120806130253_02.mat 52 | /test/images/T0006_XM_20140815110055_01.jpg /test/labels/T0006_XM_20140815110055_01.mat 53 | /test/images/T0006_XM_20140808110052_01.jpg /test/labels/T0006_XM_20140808110052_01.mat 54 | /test/images/T0006_XM_20140810160052_01.jpg /test/labels/T0006_XM_20140810160052_01.mat 55 | /test/images/T0002_XM_20110727160223_01.jpg /test/labels/T0002_XM_20110727160223_01.mat 56 | /test/images/XAM05_YM_20150729160225_01.jpg /test/labels/XAM05_YM_20150729160225_01.mat 57 | /test/images/T0001_XM_20110813100247_01.jpg /test/labels/T0001_XM_20110813100247_01.mat 58 | /test/images/T0001_YM_20100805140235_01.jpg /test/labels/T0001_YM_20100805140235_01.mat 59 | /test/images/T0002_XM_20110730090225_01.jpg /test/labels/T0002_XM_20110730090225_01.mat 60 | /test/images/T0002_XM_20110728150223_01.jpg /test/labels/T0002_XM_20110728150223_01.mat 61 | /test/images/T0001_XM_20120805100258_02.jpg /test/labels/T0001_XM_20120805100258_02.mat 62 | /test/images/T0001_XM_20120804100258_02.jpg /test/labels/T0001_XM_20120804100258_02.mat 63 | /test/images/T0001_XM_20110809160242_01.jpg /test/labels/T0001_XM_20110809160242_01.mat 64 | /test/images/T0001_XM_20120803100257_02.jpg /test/labels/T0001_XM_20120803100257_02.mat 65 | /test/images/T0001_YM_20100805120236_01.jpg /test/labels/T0001_YM_20100805120236_01.mat 66 | /test/images/T0001_XM_20110809100242_01.jpg /test/labels/T0001_XM_20110809100242_01.mat 67 | /test/images/T0002_XM_20110802160225_01.jpg /test/labels/T0002_XM_20110802160225_01.mat 68 | /test/images/T0006_YM_20140816100050_01.jpg /test/labels/T0006_YM_20140816100050_01.mat 69 | /test/images/T0001_YM_20100806130240_01.jpg /test/labels/T0001_YM_20100806130240_01.mat 70 | /test/images/T0001_YM_20100805150240_01.jpg /test/labels/T0001_YM_20100805150240_01.mat 71 | /test/images/XAM02_YM_20150729100005_01.jpg /test/labels/XAM02_YM_20150729100005_01.mat 72 | /test/images/T0001_XM_20120803090254_02.jpg /test/labels/T0001_XM_20120803090254_02.mat 73 | /test/images/T0006_XM_20140814090050_01.jpg /test/labels/T0006_XM_20140814090050_01.mat 74 | /test/images/T0001_XM_20120804160257_02.jpg /test/labels/T0001_XM_20120804160257_02.mat 75 | /test/images/T0001_YM_20100802160235_01.jpg /test/labels/T0001_YM_20100802160235_01.mat 76 | /test/images/T0001_YM_20100803100238_01.jpg /test/labels/T0001_YM_20100803100238_01.mat 77 | /test/images/XAM05_YM_20150724160229_01.jpg /test/labels/XAM05_YM_20150724160229_01.mat 78 | /test/images/T0006_XM_20140815180047_01.jpg /test/labels/T0006_XM_20140815180047_01.mat 79 | /test/images/T0001_XM_20110811120244_01.jpg /test/labels/T0001_XM_20110811120244_01.mat 80 | /test/images/XAM05_YM_20150723160224_01.jpg /test/labels/XAM05_YM_20150723160224_01.mat 81 | /test/images/T0001_YM_20100807100236_01.jpg /test/labels/T0001_YM_20100807100236_01.mat 82 | /test/images/T0002_XM_20110730130223_01.jpg /test/labels/T0002_XM_20110730130223_01.mat 83 | /test/images/T0002_XM_20110801150224_01.jpg /test/labels/T0002_XM_20110801150224_01.mat 84 | /test/images/T0001_XM_20120805140252_02.jpg /test/labels/T0001_XM_20120805140252_02.mat 85 | /test/images/T0006_XM_20140812110051_01.jpg /test/labels/T0006_XM_20140812110051_01.mat 86 | /test/images/T0006_XM_20140814110050_01.jpg /test/labels/T0006_XM_20140814110050_01.mat 87 | /test/images/T0001_XM_20130805130253_02.jpg /test/labels/T0001_XM_20130805130253_02.mat 88 | /test/images/T0002_XM_20110802110223_01.jpg /test/labels/T0002_XM_20110802110223_01.mat 89 | /test/images/T0001_XM_20110807100242_01.jpg /test/labels/T0001_XM_20110807100242_01.mat 90 | /test/images/T0001_YM_20100807090237_01.jpg /test/labels/T0001_YM_20100807090237_01.mat 91 | /test/images/T0006_XM_20140809180047_01.jpg /test/labels/T0006_XM_20140809180047_01.mat 92 | /test/images/T0001_XM_20130804080253_02.jpg /test/labels/T0001_XM_20130804080253_02.mat 93 | /test/images/T0002_XM_20110730160224_01.jpg /test/labels/T0002_XM_20110730160224_01.mat 94 | /test/images/T0006_XM_20140814150051_01.jpg /test/labels/T0006_XM_20140814150051_01.mat 95 | /test/images/T0001_YM_20100809160239_01.jpg /test/labels/T0001_YM_20100809160239_01.mat 96 | /test/images/T0001_XM_20110812160245_01.jpg /test/labels/T0001_XM_20110812160245_01.mat 97 | /test/images/T0001_YM_20100806110235_01.jpg /test/labels/T0001_YM_20100806110235_01.mat 98 | /test/images/T0001_XM_20130808110252_02.jpg /test/labels/T0001_XM_20130808110252_02.mat 99 | /test/images/T0006_XM_20140809100048_01.jpg /test/labels/T0006_XM_20140809100048_01.mat 100 | /test/images/T0001_YM_20100806160239_01.jpg /test/labels/T0001_YM_20100806160239_01.mat 101 | /test/images/XAM02_YM_20150802100007_01.jpg /test/labels/XAM02_YM_20150802100007_01.mat 102 | /test/images/XAM02_YM_20150805100009_01.jpg /test/labels/XAM02_YM_20150805100009_01.mat 103 | /test/images/T0001_XM_20120803120254_02.jpg /test/labels/T0001_XM_20120803120254_02.mat 104 | /test/images/T0002_XM_20110730110224_01.jpg /test/labels/T0002_XM_20110730110224_01.mat 105 | /test/images/T0001_YM_20100806150239_01.jpg /test/labels/T0001_YM_20100806150239_01.mat 106 | /test/images/T0001_YM_20100805100237_01.jpg /test/labels/T0001_YM_20100805100237_01.mat 107 | /test/images/XAM02_YM_20150804100010_01.jpg /test/labels/XAM02_YM_20150804100010_01.mat 108 | /test/images/T0002_XM_20110731100223_01.jpg /test/labels/T0002_XM_20110731100223_01.mat 109 | /test/images/T0001_XM_20120802160252_02.jpg /test/labels/T0001_XM_20120802160252_02.mat 110 | /test/images/T0001_XM_20120805130255_02.jpg /test/labels/T0001_XM_20120805130255_02.mat 111 | /test/images/T0001_XM_20130806120253_02.jpg /test/labels/T0001_XM_20130806120253_02.mat 112 | /test/images/T0002_XM_20110801160224_01.jpg /test/labels/T0002_XM_20110801160224_01.mat 113 | /test/images/T0001_XM_20130809120253_02.jpg /test/labels/T0001_XM_20130809120253_02.mat 114 | /test/images/T0006_YM_20140817160050_01.jpg /test/labels/T0006_YM_20140817160050_01.mat 115 | /test/images/XAM02_YM_20150803100005_01.jpg /test/labels/XAM02_YM_20150803100005_01.mat 116 | /test/images/XAM02_YM_20150801100008_01.jpg /test/labels/XAM02_YM_20150801100008_01.mat 117 | /test/images/T0006_XM_20140808160048_01.jpg /test/labels/T0006_XM_20140808160048_01.mat 118 | /test/images/T0001_YM_20100807160235_01.jpg /test/labels/T0001_YM_20100807160235_01.mat 119 | /test/images/XAM02_YM_20150726100009_01.jpg /test/labels/XAM02_YM_20150726100009_01.mat 120 | /test/images/T0002_XM_20110730100225_01.jpg /test/labels/T0002_XM_20110730100225_01.mat 121 | /test/images/T0001_YM_20100809130240_01.jpg /test/labels/T0001_YM_20100809130240_01.mat 122 | /test/images/T0006_XM_20140811100052_01.jpg /test/labels/T0006_XM_20140811100052_01.mat 123 | /test/images/XAM05_YM_20150802160229_01.jpg /test/labels/XAM05_YM_20150802160229_01.mat 124 | /test/images/T0001_YM_20100809100235_01.jpg /test/labels/T0001_YM_20100809100235_01.mat 125 | /test/images/T0001_YM_20100809140240_01.jpg /test/labels/T0001_YM_20100809140240_01.mat 126 | /test/images/XAM05_YM_20150803160225_01.jpg /test/labels/XAM05_YM_20150803160225_01.mat 127 | /test/images/T0001_YM_20100806100235_01.jpg /test/labels/T0001_YM_20100806100235_01.mat 128 | /test/images/T0001_YM_20100808100235_01.jpg /test/labels/T0001_YM_20100808100235_01.mat 129 | /test/images/T0006_XM_20140810180047_01.jpg /test/labels/T0006_XM_20140810180047_01.mat 130 | /test/images/T0006_XM_20140811130049_01.jpg /test/labels/T0006_XM_20140811130049_01.mat 131 | /test/images/T0001_YM_20100802100241_01.jpg /test/labels/T0001_YM_20100802100241_01.mat 132 | /test/images/T0002_XM_20110729110232_01.jpg /test/labels/T0002_XM_20110729110232_01.mat 133 | /test/images/T0001_XM_20110810160246_01.jpg /test/labels/T0001_XM_20110810160246_01.mat 134 | /test/images/T0001_XM_20110810100242_01.jpg /test/labels/T0001_XM_20110810100242_01.mat 135 | /test/images/T0001_YM_20100804160240_01.jpg /test/labels/T0001_YM_20100804160240_01.mat 136 | /test/images/T0001_YM_20100806140240_01.jpg /test/labels/T0001_YM_20100806140240_01.mat 137 | /test/images/T0006_XM_20140812140053_01.jpg /test/labels/T0006_XM_20140812140053_01.mat 138 | /test/images/T0001_XM_20110811100244_01.jpg /test/labels/T0001_XM_20110811100244_01.mat 139 | /test/images/T0006_YM_20140816160047_01.jpg /test/labels/T0006_YM_20140816160047_01.mat 140 | /test/images/T0001_XM_20120805110255_02.jpg /test/labels/T0001_XM_20120805110255_02.mat 141 | /test/images/T0001_YM_20100806090236_01.jpg /test/labels/T0001_YM_20100806090236_01.mat 142 | /test/images/T0001_XM_20120805120252_02.jpg /test/labels/T0001_XM_20120805120252_02.mat 143 | /test/images/T0001_YM_20100807110237_01.jpg /test/labels/T0001_YM_20100807110237_01.mat 144 | /test/images/T0001_XM_20110813090243_01.jpg /test/labels/T0001_XM_20110813090243_01.mat 145 | /test/images/T0006_XM_20140812170051_01.jpg /test/labels/T0006_XM_20140812170051_01.mat 146 | /test/images/T0001_XM_20110811140245_01.jpg /test/labels/T0001_XM_20110811140245_01.mat 147 | /test/images/T0006_XM_20140810150050_01.jpg /test/labels/T0006_XM_20140810150050_01.mat 148 | /test/images/XAM02_YM_20150728100006_01.jpg /test/labels/XAM02_YM_20150728100006_01.mat 149 | /test/images/T0001_XM_20120807150255_02.jpg /test/labels/T0001_XM_20120807150255_02.mat 150 | /test/images/XAM05_YM_20150730160225_01.jpg /test/labels/XAM05_YM_20150730160225_01.mat 151 | /test/images/T0001_YM_20100806120242_01.jpg /test/labels/T0001_YM_20100806120242_01.mat 152 | /test/images/T0001_YM_20100805130235_01.jpg /test/labels/T0001_YM_20100805130235_01.mat 153 | /test/images/XAM05_YM_20150727160224_01.jpg /test/labels/XAM05_YM_20150727160224_01.mat 154 | /test/images/T0006_YM_20140818130051_01.jpg /test/labels/T0006_YM_20140818130051_01.mat 155 | /test/images/T0006_YM_20140817130050_01.jpg /test/labels/T0006_YM_20140817130050_01.mat 156 | /test/images/XAM02_YM_20150730100009_01.jpg /test/labels/XAM02_YM_20150730100009_01.mat 157 | /test/images/T0002_XM_20110801090223_01.jpg /test/labels/T0002_XM_20110801090223_01.mat 158 | /test/images/T0002_XM_20110801100224_01.jpg /test/labels/T0002_XM_20110801100224_01.mat 159 | /test/images/T0006_XM_20140810120053_01.jpg /test/labels/T0006_XM_20140810120053_01.mat 160 | /test/images/T0001_YM_20100805160240_01.jpg /test/labels/T0001_YM_20100805160240_01.mat 161 | /test/images/T0001_XM_20110812130242_01.jpg /test/labels/T0001_XM_20110812130242_01.mat 162 | /test/images/T0006_XM_20140813140046_01.jpg /test/labels/T0006_XM_20140813140046_01.mat 163 | /test/images/T0006_XM_20140808140052_01.jpg /test/labels/T0006_XM_20140808140052_01.mat 164 | /test/images/T0006_XM_20140809140048_01.jpg /test/labels/T0006_XM_20140809140048_01.mat 165 | /test/images/T0001_XM_20130807100253_02.jpg /test/labels/T0001_XM_20130807100253_02.mat 166 | /test/images/T0001_XM_20130803160252_02.jpg /test/labels/T0001_XM_20130803160252_02.mat 167 | /test/images/T0006_XM_20140810090049_01.jpg /test/labels/T0006_XM_20140810090049_01.mat 168 | /test/images/T0001_XM_20120805160256_02.jpg /test/labels/T0001_XM_20120805160256_02.mat 169 | /test/images/T0001_YM_20100808130241_01.jpg /test/labels/T0001_YM_20100808130241_01.mat 170 | /test/images/T0006_XM_20140809160052_01.jpg /test/labels/T0006_XM_20140809160052_01.mat 171 | /test/images/T0002_XM_20110731160223_01.jpg /test/labels/T0002_XM_20110731160223_01.mat 172 | /test/images/XAM05_YM_20150728150224_01.jpg /test/labels/XAM05_YM_20150728150224_01.mat 173 | /test/images/T0006_XM_20140815140051_01.jpg /test/labels/T0006_XM_20140815140051_01.mat 174 | /test/images/T0006_YM_20140818150054_01.jpg /test/labels/T0006_YM_20140818150054_01.mat 175 | /test/images/T0001_XM_20120803160256_02.jpg /test/labels/T0001_XM_20120803160256_02.mat -------------------------------------------------------------------------------- /hldataset.py: -------------------------------------------------------------------------------- 1 | """ 2 | @author: hao lu 3 | """ 4 | 5 | import os 6 | import json 7 | import matplotlib.pyplot as plt 8 | import matplotlib.cm as cm 9 | import pandas as pd 10 | import random 11 | import numpy as np 12 | from PIL import Image 13 | import cv2 14 | import h5py 15 | import scipy.io as sio 16 | from scipy.ndimage.filters import gaussian_filter 17 | from skimage import util 18 | from skimage.measure import label 19 | from skimage.measure import regionprops 20 | import xml.etree.ElementTree as ET 21 | 22 | import torch 23 | from torch.utils.data import Dataset 24 | from torchvision import transforms 25 | import torch.nn.functional as F 26 | 27 | 28 | def read_image(x): 29 | img_arr = np.array(Image.open(x)) 30 | if len(img_arr.shape) == 2: # grayscale 31 | img_arr = np.tile(img_arr, [3, 1, 1]).transpose(1, 2, 0) 32 | return img_arr 33 | 34 | 35 | class RandomCrop(object): 36 | def __init__(self, output_size): 37 | assert isinstance(output_size, (int, tuple)) 38 | self.output_size = output_size 39 | 40 | def __call__(self, sample): 41 | 42 | image, target, gtcount = sample['image'], sample['target'], sample['gtcount'] 43 | h, w = image.shape[:2] 44 | 45 | if isinstance(self.output_size, tuple): 46 | new_h = min(self.output_size[0], h) 47 | new_w = min(self.output_size[1], w) 48 | assert (new_h, new_w) == self.output_size 49 | else: 50 | crop_size = min(self.output_size, h, w) 51 | assert crop_size == self.output_size 52 | new_h = new_w = crop_size 53 | if gtcount > 0: 54 | mask = target > 0 55 | ch, cw = int(np.ceil(new_h / 2)), int(np.ceil(new_w / 2)) 56 | mask_center = np.zeros((h, w), dtype=np.uint8) 57 | mask_center[ch:h-ch+1, cw:w-cw+1] = 1 58 | mask = (mask & mask_center) 59 | idh, idw = np.where(mask == 1) 60 | if len(idh) != 0: 61 | ids = random.choice(range(len(idh))) 62 | hc, wc = idh[ids], idw[ids] 63 | top, left = hc-ch, wc-cw 64 | else: 65 | top = np.random.randint(0, h-new_h+1) 66 | left = np.random.randint(0, w-new_w+1) 67 | else: 68 | top = np.random.randint(0, h-new_h+1) 69 | left = np.random.randint(0, w-new_w+1) 70 | 71 | image = image[top:top+new_h, left:left+new_w, :] 72 | target = target[top:top+new_h, left:left+new_w] 73 | 74 | return {'image': image, 'target': target, 'gtcount': gtcount} 75 | 76 | 77 | class RandomFlip(object): 78 | def __init__(self): 79 | pass 80 | 81 | def __call__(self, sample): 82 | image, target, gtcount = sample['image'], sample['target'], sample['gtcount'] 83 | do_mirror = np.random.randint(2) 84 | if do_mirror: 85 | image = cv2.flip(image, 1) 86 | target = cv2.flip(target, 1) 87 | return {'image': image, 'target': target, 'gtcount': gtcount} 88 | 89 | 90 | class Normalize(object): 91 | 92 | def __init__(self, scale, mean, std): 93 | self.scale = scale 94 | self.mean = mean 95 | self.std = std 96 | 97 | def __call__(self, sample): 98 | image, target, gtcount = sample['image'], sample['target'], sample['gtcount'] 99 | image, target = image.astype('float32'), target.astype('float32') 100 | 101 | # pixel normalization 102 | image = (self.scale * image - self.mean) / self.std 103 | 104 | image, target = image.astype('float32'), target.astype('float32') 105 | 106 | return {'image': image, 'target': target, 'gtcount': gtcount} 107 | 108 | 109 | class ZeroPadding(object): 110 | def __init__(self, psize=32): 111 | self.psize = psize 112 | 113 | def __call__(self, sample): 114 | psize = self.psize 115 | 116 | image, target, gtcount = sample['image'], sample['target'], sample['gtcount'] 117 | h,w = image.size()[-2:] 118 | ph,pw = (psize-h%psize),(psize-w%psize) 119 | # print(ph,pw) 120 | 121 | (pl, pr) = (pw//2, pw-pw//2) if pw != psize else (0, 0) 122 | (pt, pb) = (ph//2, ph-ph//2) if ph != psize else (0, 0) 123 | if (ph!=psize) or (pw!=psize): 124 | tmp_pad = [pl, pr, pt, pb] 125 | # print(tmp_pad) 126 | image = F.pad(image,tmp_pad) 127 | target = F.pad(target,tmp_pad) 128 | 129 | return {'image': image, 'target': target, 'gtcount': gtcount} 130 | 131 | 132 | class ToTensor(object): 133 | """Convert ndarrays in sample to Tensors.""" 134 | 135 | def __init__(self): 136 | pass 137 | 138 | def __call__(self, sample): 139 | # swap color axis 140 | # numpy image: H x W x C 141 | # torch image: C X H X W 142 | image, target, gtcount = sample['image'], sample['target'], sample['gtcount'] 143 | image = image.transpose((2, 0, 1)) 144 | target = np.expand_dims(target, axis=2) 145 | target = target.transpose((2, 0, 1)) 146 | image, target = torch.from_numpy(image), torch.from_numpy(target) 147 | return {'image': image, 'target': target, 'gtcount': gtcount} 148 | 149 | 150 | class MaizeTasselDataset(Dataset): 151 | def __init__(self, data_dir, data_list, ratio, train=True, transform=None): 152 | self.data_dir = data_dir 153 | self.data_list = [name.split('\t') for name in open(data_list).read().splitlines()] 154 | self.ratio = ratio 155 | self.train = train 156 | self.transform = transform 157 | self.image_list = [] 158 | 159 | # store images and generate ground truths 160 | self.images = {} 161 | self.targets = {} 162 | self.gtcounts = {} 163 | self.dotimages = {} 164 | 165 | def bbs2points(self, bbs): 166 | points = [] 167 | for bb in bbs: 168 | x1, y1, w, h = [float(b) for b in bb] 169 | x2, y2 = x1+w-1, y1+h-1 170 | x, y = np.round((x1+x2)/2).astype(np.int32), np.round((y1+y2)/2).astype(np.int32) 171 | points.append([x, y]) 172 | return points 173 | 174 | def __len__(self): 175 | return len(self.data_list) 176 | 177 | def __getitem__(self, idx): 178 | file_name = self.data_list[idx] 179 | self.image_list.append(file_name[0]) 180 | if file_name[0] not in self.images: 181 | image = read_image(self.data_dir+file_name[0]) 182 | annotation = sio.loadmat(self.data_dir+file_name[1]) 183 | h, w = image.shape[:2] 184 | nh = int(np.ceil(h * self.ratio)) 185 | nw = int(np.ceil(w * self.ratio)) 186 | image = cv2.resize(image, (nw, nh), interpolation = cv2.INTER_CUBIC) 187 | target = np.zeros((nh, nw), dtype=np.float32) 188 | dotimage = image.copy() 189 | if annotation['annotation'][0][0][1] is not None: 190 | bbs = annotation['annotation'][0][0][1] 191 | gtcount = bbs.shape[0] 192 | pts = self.bbs2points(bbs) 193 | for pt in pts: 194 | pt[0], pt[1] = int(pt[0] * self.ratio), int(pt[1] * self.ratio) 195 | target[pt[1], pt[0]] = 1 196 | cv2.circle(dotimage, (pt[0], pt[1]), int(24 * self.ratio) , (255, 0, 0), -1) 197 | else: 198 | gtcount = 0 199 | target = gaussian_filter(target, 80 * self.ratio) 200 | 201 | # plt.imshow(target, cmap=cm.jet) 202 | # plt.show() 203 | # print(target.sum()) 204 | 205 | self.images.update({file_name[0]:image}) 206 | self.targets.update({file_name[0]:target}) 207 | self.gtcounts.update({file_name[0]:gtcount}) 208 | self.dotimages.update({file_name[0]:dotimage}) 209 | 210 | 211 | sample = { 212 | 'image': self.images[file_name[0]], 213 | 'target': self.targets[file_name[0]], 214 | 'gtcount': self.gtcounts[file_name[0]] 215 | } 216 | 217 | if self.transform: 218 | sample = self.transform(sample) 219 | 220 | return sample 221 | 222 | 223 | class WhearEarDataset(Dataset): 224 | def __init__(self, data_dir, data_list, ratio, train=True, transform=None): 225 | self.data_dir = data_dir 226 | self.data_list = [name.split('\t') for name in open(data_list).read().splitlines()] 227 | self.ratio = ratio 228 | self.train = train 229 | self.transform = transform 230 | self.image_list = [] 231 | 232 | # store images and generate ground truths 233 | self.images = {} 234 | self.targets = {} 235 | self.gtcounts = {} 236 | self.dotimages = {} 237 | 238 | def parsexml(self, xml): 239 | tree = ET.parse(xml) 240 | root = tree.getroot() 241 | bbs = [] 242 | for bb in root.iter('bndbox'): 243 | xmin = int(bb.find('xmin').text) 244 | ymin = int(bb.find('ymin').text) 245 | xmax = int(bb.find('xmax').text) 246 | ymax = int(bb.find('ymax').text) 247 | bbs.append([xmin, ymin, xmax, ymax]) 248 | return bbs 249 | 250 | def bbs2points(self, bbs): 251 | points = [] 252 | for bb in bbs: 253 | x1, y1, x2, y2 = [float(b) for b in bb] 254 | x, y = np.round((x1+x2)/2).astype(np.int32), np.round((y1+y2)/2).astype(np.int32) 255 | points.append([x, y]) 256 | return points 257 | 258 | def __len__(self): 259 | return len(self.data_list) 260 | 261 | def __getitem__(self, idx): 262 | file_name = self.data_list[idx] 263 | self.image_list.append(file_name[0]) 264 | if file_name[0] not in self.images: 265 | image = read_image(self.data_dir+file_name[0]) 266 | bbs = self.parsexml(self.data_dir+file_name[1]) 267 | h, w = image.shape[:2] 268 | nh = int(np.ceil(h * self.ratio)) 269 | nw = int(np.ceil(w * self.ratio)) 270 | image = cv2.resize(image, (nw, nh), interpolation = cv2.INTER_CUBIC) 271 | target = np.zeros((nh, nw), dtype=np.float32) 272 | dotimage = image.copy() 273 | if bbs is not None: 274 | gtcount = len(bbs) 275 | pts = self.bbs2points(bbs) 276 | for pt in pts: 277 | pt[0], pt[1] = int(pt[0] * self.ratio), int(pt[1] * self.ratio) 278 | target[pt[1], pt[0]] = 1 279 | cv2.circle(dotimage, (pt[0], pt[1]), 6, (255, 0, 0), -1) 280 | else: 281 | gtcount = 0 282 | target = gaussian_filter(target, 40 * self.ratio) 283 | 284 | # cmap = plt.cm.get_cmap('jet') 285 | # target_show = target / (target.max() + 1e-12) 286 | # target_show = cmap(target_show) * 255. 287 | # target_show = 0.5 * image + 0.5 * target_show[:, :, 0:3] 288 | # plt.imshow(target_show.astype(np.uint8)) 289 | # plt.show() 290 | # print(target.sum()) 291 | 292 | # plt.imshow(dotimage.astype(np.uint8)) 293 | # plt.show() 294 | 295 | self.images.update({file_name[0]:image}) 296 | self.targets.update({file_name[0]:target}) 297 | self.gtcounts.update({file_name[0]:gtcount}) 298 | self.dotimages.update({file_name[0]:dotimage}) 299 | 300 | 301 | sample = { 302 | 'image': self.images[file_name[0]], 303 | 'target': self.targets[file_name[0]], 304 | 'gtcount': self.gtcounts[file_name[0]] 305 | } 306 | 307 | if self.transform: 308 | sample = self.transform(sample) 309 | 310 | return sample 311 | 312 | 313 | class SorghumHeadDataset(Dataset): 314 | def __init__(self, data_dir, data_list, ratio, train=True, transform=None): 315 | self.data_dir = data_dir 316 | self.data_list = [name.split('\t') for name in open(data_list).read().splitlines()] 317 | self.ratio = ratio 318 | self.train = train 319 | self.transform = transform 320 | self.image_list = [] 321 | 322 | # store images and generate ground truths 323 | self.images = {} 324 | self.targets = {} 325 | self.gtcounts = {} 326 | self.dotimages = {} 327 | 328 | def __len__(self): 329 | return len(self.data_list) 330 | 331 | def __getitem__(self, idx): 332 | file_name = self.data_list[idx] 333 | self.image_list.append(file_name[0]) 334 | if file_name[0] not in self.images: 335 | image = read_image(self.data_dir+file_name[0]) 336 | annotations = read_image(self.data_dir+file_name[1]) 337 | annotations = util.img_as_ubyte(annotations[:, :, 0]) == 0 338 | annotations = label(annotations, connectivity=annotations.ndim) 339 | annotations = regionprops(annotations) 340 | 341 | h, w = image.shape[:2] 342 | nh = int(np.ceil(h * self.ratio)) 343 | nw = int(np.ceil(w * self.ratio)) 344 | image = cv2.resize(image, (nw, nh), interpolation = cv2.INTER_CUBIC) 345 | target = np.zeros((nh, nw), dtype=np.float32) 346 | dotimage = image.copy() 347 | if annotations is not None: 348 | gtcount = len(annotations) 349 | for annotation in annotations: 350 | pt = annotation.centroid 351 | x, y = int(pt[0] * self.ratio), int(pt[1] * self.ratio) 352 | target[x, y] = 1 353 | cv2.circle(dotimage, (y, x), 6, (255, 0, 0), -1) 354 | else: 355 | gtcount = 0 356 | target = gaussian_filter(target, 8 * self.ratio) 357 | 358 | # cmap = plt.cm.get_cmap('jet') 359 | # target_show = target / (target.max() + 1e-12) 360 | # target_show = cmap(target_show) * 255. 361 | # target_show = 0.5 * image + 0.5 * target_show[:, :, 0:3] 362 | # plt.imshow(target_show.astype(np.uint8)) 363 | # plt.show() 364 | # print(target.sum()) 365 | 366 | # plt.imshow(dotimage.astype(np.uint8)) 367 | # plt.show() 368 | 369 | self.images.update({file_name[0]:image}) 370 | self.targets.update({file_name[0]:target}) 371 | self.gtcounts.update({file_name[0]:gtcount}) 372 | self.dotimages.update({file_name[0]:dotimage}) 373 | 374 | 375 | sample = { 376 | 'image': self.images[file_name[0]], 377 | 'target': self.targets[file_name[0]], 378 | 'gtcount': self.gtcounts[file_name[0]] 379 | } 380 | 381 | if self.transform: 382 | sample = self.transform(sample) 383 | 384 | return sample 385 | 386 | 387 | if __name__=='__main__': 388 | 389 | dataset = WhearEarDataset( 390 | data_dir='./data/wheat_ears_counting_dataset', 391 | data_list='./data/wheat_ears_counting_dataset/train.txt', 392 | ratio=0.167, 393 | train=True, 394 | transform=transforms.Compose([ 395 | ToTensor()] 396 | ) 397 | ) 398 | 399 | dataloader = torch.utils.data.DataLoader( 400 | dataset, 401 | batch_size=1, 402 | shuffle=False, 403 | num_workers=0 404 | ) 405 | 406 | print(len(dataloader)) 407 | mean = 0. 408 | std = 0. 409 | for i, data in enumerate(dataloader, 0): 410 | images, targets = data['image'], data['target'] 411 | bs = images.size(0) 412 | images = images.view(bs, images.size(1), -1).float() 413 | mean += images.mean(2).sum(0) 414 | std += images.std(2).sum(0) 415 | print(images.size()) 416 | print(i) 417 | mean /= len(dataloader) 418 | std /= len(dataloader) 419 | print(mean/255.) 420 | print(std/255.) -------------------------------------------------------------------------------- /data/maize_counting_dataset/train.txt: -------------------------------------------------------------------------------- 1 | /trainval/images/T0002_XM_20120805160221_01.jpg /trainval/labels/T0002_XM_20120805160221_01.mat 2 | /trainval/images/T0001_XM_20120805120252_01.jpg /trainval/labels/T0001_XM_20120805120252_01.mat 3 | /trainval/images/T0001_XM_20110811160246_02.jpg /trainval/labels/T0001_XM_20110811160246_02.mat 4 | /trainval/images/T0002_YM_20100817110218_01.jpg /trainval/labels/T0002_YM_20100817110218_01.mat 5 | /trainval/images/T0001_XM_20110812160246_02.jpg /trainval/labels/T0001_XM_20110812160246_02.mat 6 | /trainval/images/XAM01_YM_20150802100255_01.jpg /trainval/labels/XAM01_YM_20150802100255_01.mat 7 | /trainval/images/T0001_XM_20110814130246_02.jpg /trainval/labels/T0001_XM_20110814130246_02.mat 8 | /trainval/images/XAM01_YM_20150804100259_01.jpg /trainval/labels/XAM01_YM_20150804100259_01.mat 9 | /trainval/images/T0001_XM_20110812100244_02.jpg /trainval/labels/T0001_XM_20110812100244_02.mat 10 | /trainval/images/T0002_XM_20120802160222_01.jpg /trainval/labels/T0002_XM_20120802160222_01.mat 11 | /trainval/images/T0001_YM_20100809160240_02.jpg /trainval/labels/T0001_YM_20100809160240_02.mat 12 | /trainval/images/T0002_YM_20100820100221_01.jpg /trainval/labels/T0002_YM_20100820100221_01.mat 13 | /trainval/images/T0001_YM_20100807090237_02.jpg /trainval/labels/T0001_YM_20100807090237_02.mat 14 | /trainval/images/T0002_YM_20100815160218_01.jpg /trainval/labels/T0002_YM_20100815160218_01.mat 15 | /trainval/images/T0001_YM_20100803100239_02.jpg /trainval/labels/T0001_YM_20100803100239_02.mat 16 | /trainval/images/T0001_XM_20120805130255_01.jpg /trainval/labels/T0001_XM_20120805130255_01.mat 17 | /trainval/images/T0002_YM_20100817090218_01.jpg /trainval/labels/T0002_YM_20100817090218_01.mat 18 | /trainval/images/T0001_XM_20120807120256_01.jpg /trainval/labels/T0001_XM_20120807120256_01.mat 19 | /trainval/images/T0001_YM_20100808110236_02.jpg /trainval/labels/T0001_YM_20100808110236_02.mat 20 | /trainval/images/T0001_XM_20120809130255_01.jpg /trainval/labels/T0001_XM_20120809130255_01.mat 21 | /trainval/images/T0001_XM_20120808140252_01.jpg /trainval/labels/T0001_XM_20120808140252_01.mat 22 | /trainval/images/T0002_YM_20100816100220_01.jpg /trainval/labels/T0002_YM_20100816100220_01.mat 23 | /trainval/images/T0002_XM_20120806090222_01.jpg /trainval/labels/T0002_XM_20120806090222_01.mat 24 | /trainval/images/T0002_YM_20100811130220_01.jpg /trainval/labels/T0002_YM_20100811130220_01.mat 25 | /trainval/images/T0001_XM_20120803100255_01.jpg /trainval/labels/T0001_XM_20120803100255_01.mat 26 | /trainval/images/T0001_XM_20110813090243_02.jpg /trainval/labels/T0001_XM_20110813090243_02.mat 27 | /trainval/images/T0002_YM_20100809110218_01.jpg /trainval/labels/T0002_YM_20100809110218_01.mat 28 | /trainval/images/T0001_YM_20100802160236_02.jpg /trainval/labels/T0001_YM_20100802160236_02.mat 29 | /trainval/images/T0001_YM_20100808140237_02.jpg /trainval/labels/T0001_YM_20100808140237_02.mat 30 | /trainval/images/T0002_XM_20120804140223_01.jpg /trainval/labels/T0002_XM_20120804140223_01.mat 31 | /trainval/images/T0001_XM_20130807100253_01.jpg /trainval/labels/T0001_XM_20130807100253_01.mat 32 | /trainval/images/T0006_XM_20120815100308_01.jpg /trainval/labels/T0006_XM_20120815100308_01.mat 33 | /trainval/images/T0001_XM_20110811090244_02.jpg /trainval/labels/T0001_XM_20110811090244_02.mat 34 | /trainval/images/T0002_YM_20100814100220_01.jpg /trainval/labels/T0002_YM_20100814100220_01.mat 35 | /trainval/images/T0002_YM_20100816160218_01.jpg /trainval/labels/T0002_YM_20100816160218_01.mat 36 | /trainval/images/T0001_XM_20110809150249_02.jpg /trainval/labels/T0001_XM_20110809150249_02.mat 37 | /trainval/images/T0002_XM_20120802120222_01.jpg /trainval/labels/T0002_XM_20120802120222_01.mat 38 | /trainval/images/T0002_XM_20120805100222_01.jpg /trainval/labels/T0002_XM_20120805100222_01.mat 39 | /trainval/images/T0001_XM_20110813100248_02.jpg /trainval/labels/T0001_XM_20110813100248_02.mat 40 | /trainval/images/T0002_YM_20100823100222_01.jpg /trainval/labels/T0002_YM_20100823100222_01.mat 41 | /trainval/images/T0001_XM_20120804140256_01.jpg /trainval/labels/T0001_XM_20120804140256_01.mat 42 | /trainval/images/T0002_XM_20120802110223_01.jpg /trainval/labels/T0002_XM_20120802110223_01.mat 43 | /trainval/images/T0001_XM_20130803160251_01.jpg /trainval/labels/T0001_XM_20130803160251_01.mat 44 | /trainval/images/T0002_YM_20100730140218_01.jpg /trainval/labels/T0002_YM_20100730140218_01.mat 45 | /trainval/images/T0001_XM_20120803130257_01.jpg /trainval/labels/T0001_XM_20120803130257_01.mat 46 | /trainval/images/T0002_XM_20120804100222_01.jpg /trainval/labels/T0002_XM_20120804100222_01.mat 47 | /trainval/images/T0001_YM_20100806130242_02.jpg /trainval/labels/T0001_YM_20100806130242_02.mat 48 | /trainval/images/XAM01_YM_20150728100255_01.jpg /trainval/labels/XAM01_YM_20150728100255_01.mat 49 | /trainval/images/T0001_XM_20120809110250_01.jpg /trainval/labels/T0001_XM_20120809110250_01.mat 50 | /trainval/images/T0001_XM_20120805160256_01.jpg /trainval/labels/T0001_XM_20120805160256_01.mat 51 | /trainval/images/T0002_YM_20100731120218_01.jpg /trainval/labels/T0002_YM_20100731120218_01.mat 52 | /trainval/images/T0002_YM_20100819100220_01.jpg /trainval/labels/T0002_YM_20100819100220_01.mat 53 | /trainval/images/T0002_YM_20100808120218_01.jpg /trainval/labels/T0002_YM_20100808120218_01.mat 54 | /trainval/images/T0002_YM_20100820160219_01.jpg /trainval/labels/T0002_YM_20100820160219_01.mat 55 | /trainval/images/T0001_XM_20120805110254_01.jpg /trainval/labels/T0001_XM_20120805110254_01.mat 56 | /trainval/images/T0002_YM_20100818100222_01.jpg /trainval/labels/T0002_YM_20100818100222_01.mat 57 | /trainval/images/T0001_XM_20120809120253_01.jpg /trainval/labels/T0001_XM_20120809120253_01.mat 58 | /trainval/images/T0001_YM_20100809100237_02.jpg /trainval/labels/T0001_YM_20100809100237_02.mat 59 | /trainval/images/T0002_XM_20120801160222_01.jpg /trainval/labels/T0002_XM_20120801160222_01.mat 60 | /trainval/images/T0001_XM_20120808120253_01.jpg /trainval/labels/T0001_XM_20120808120253_01.mat 61 | /trainval/images/T0001_XM_20120808150253_01.jpg /trainval/labels/T0001_XM_20120808150253_01.mat 62 | /trainval/images/XAM01_YM_20150801100257_01.jpg /trainval/labels/XAM01_YM_20150801100257_01.mat 63 | /trainval/images/T0002_YM_20100729100220_01.jpg /trainval/labels/T0002_YM_20100729100220_01.mat 64 | /trainval/images/T0001_XM_20120801160256_01.jpg /trainval/labels/T0001_XM_20120801160256_01.mat 65 | /trainval/images/T0002_YM_20100822090219_01.jpg /trainval/labels/T0002_YM_20100822090219_01.mat 66 | /trainval/images/T0002_YM_20100821100220_01.jpg /trainval/labels/T0002_YM_20100821100220_01.mat 67 | /trainval/images/T0001_YM_20100806120243_02.jpg /trainval/labels/T0001_YM_20100806120243_02.mat 68 | /trainval/images/XAM01_YM_20150731100257_01.jpg /trainval/labels/XAM01_YM_20150731100257_01.mat 69 | /trainval/images/T0001_XM_20120803140255_01.jpg /trainval/labels/T0001_XM_20120803140255_01.mat 70 | /trainval/images/T0002_YM_20100819090220_01.jpg /trainval/labels/T0002_YM_20100819090220_01.mat 71 | /trainval/images/T0002_XM_20120806120222_01.jpg /trainval/labels/T0002_XM_20120806120222_01.mat 72 | /trainval/images/T0001_XM_20110811100245_02.jpg /trainval/labels/T0001_XM_20110811100245_02.mat 73 | /trainval/images/T0002_XM_20120803150221_01.jpg /trainval/labels/T0002_XM_20120803150221_01.mat 74 | /trainval/images/T0001_XM_20120805150253_01.jpg /trainval/labels/T0001_XM_20120805150253_01.mat 75 | /trainval/images/T0002_YM_20100816120219_01.jpg /trainval/labels/T0002_YM_20100816120219_01.mat 76 | /trainval/images/XAM01_YM_20150729100258_01.jpg /trainval/labels/XAM01_YM_20150729100258_01.mat 77 | /trainval/images/T0001_YM_20100802100242_02.jpg /trainval/labels/T0001_YM_20100802100242_02.mat 78 | /trainval/images/T0002_YM_20100814160219_01.jpg /trainval/labels/T0002_YM_20100814160219_01.mat 79 | /trainval/images/T0002_XM_20120805130223_01.jpg /trainval/labels/T0002_XM_20120805130223_01.mat 80 | /trainval/images/T0006_XM_20120809100305_01.jpg /trainval/labels/T0006_XM_20120809100305_01.mat 81 | /trainval/images/T0002_YM_20100816140221_01.jpg /trainval/labels/T0002_YM_20100816140221_01.mat 82 | /trainval/images/T0002_YM_20100823160222_01.jpg /trainval/labels/T0002_YM_20100823160222_01.mat 83 | /trainval/images/T0001_XM_20110815120244_02.jpg /trainval/labels/T0001_XM_20110815120244_02.mat 84 | /trainval/images/T0001_XM_20110812130243_02.jpg /trainval/labels/T0001_XM_20110812130243_02.mat 85 | /trainval/images/T0001_YM_20100807100238_02.jpg /trainval/labels/T0001_YM_20100807100238_02.mat 86 | /trainval/images/T0001_XM_20120809090253_01.jpg /trainval/labels/T0001_XM_20120809090253_01.mat 87 | /trainval/images/T0001_YM_20100806160240_02.jpg /trainval/labels/T0001_YM_20100806160240_02.mat 88 | /trainval/images/T0002_XM_20120807160222_01.jpg /trainval/labels/T0002_XM_20120807160222_01.mat 89 | /trainval/images/T0002_YM_20100810160221_01.jpg /trainval/labels/T0002_YM_20100810160221_01.mat 90 | /trainval/images/T0006_XM_20120807160304_01.jpg /trainval/labels/T0006_XM_20120807160304_01.mat 91 | /trainval/images/T0001_YM_20100805140236_02.jpg /trainval/labels/T0001_YM_20100805140236_02.mat 92 | /trainval/images/T0006_XM_20120808160306_01.jpg /trainval/labels/T0006_XM_20120808160306_01.mat 93 | /trainval/images/XAM01_YM_20150726100258_01.jpg /trainval/labels/XAM01_YM_20150726100258_01.mat 94 | /trainval/images/T0002_XM_20120803090221_01.jpg /trainval/labels/T0002_XM_20120803090221_01.mat 95 | /trainval/images/T0001_XM_20120803160255_01.jpg /trainval/labels/T0001_XM_20120803160255_01.mat 96 | /trainval/images/T0001_XM_20110814160244_02.jpg /trainval/labels/T0001_XM_20110814160244_02.mat 97 | /trainval/images/XAM01_YM_20150803110256_01.jpg /trainval/labels/XAM01_YM_20150803110256_01.mat 98 | /trainval/images/T0001_YM_20100806090236_02.jpg /trainval/labels/T0001_YM_20100806090236_02.mat 99 | /trainval/images/T0002_YM_20100821090220_01.jpg /trainval/labels/T0002_YM_20100821090220_01.mat 100 | /trainval/images/T0002_YM_20100821140220_01.jpg /trainval/labels/T0002_YM_20100821140220_01.mat 101 | /trainval/images/T0001_YM_20100805150241_02.jpg /trainval/labels/T0001_YM_20100805150241_02.mat 102 | /trainval/images/T0006_XM_20120815160306_01.jpg /trainval/labels/T0006_XM_20120815160306_01.mat 103 | /trainval/images/T0006_XM_20120810140308_01.jpg /trainval/labels/T0006_XM_20120810140308_01.mat 104 | /trainval/images/T0001_YM_20100803160242_02.jpg /trainval/labels/T0001_YM_20100803160242_02.mat 105 | /trainval/images/T0001_YM_20100804100237_02.jpg /trainval/labels/T0001_YM_20100804100237_02.mat 106 | /trainval/images/T0002_XM_20120801100222_01.jpg /trainval/labels/T0002_XM_20120801100222_01.mat 107 | /trainval/images/T0002_YM_20100815100220_01.jpg /trainval/labels/T0002_YM_20100815100220_01.mat 108 | /trainval/images/T0002_XM_20120803160221_01.jpg /trainval/labels/T0002_XM_20120803160221_01.mat 109 | /trainval/images/T0006_XM_20120813100305_01.jpg /trainval/labels/T0006_XM_20120813100305_01.mat 110 | /trainval/images/T0006_XM_20120809130643_01.jpg /trainval/labels/T0006_XM_20120809130643_01.mat 111 | /trainval/images/T0001_XM_20120806130252_01.jpg /trainval/labels/T0001_XM_20120806130252_01.mat 112 | /trainval/images/T0001_YM_20100805100238_02.jpg /trainval/labels/T0001_YM_20100805100238_02.mat 113 | /trainval/images/T0002_YM_20100808140219_01.jpg /trainval/labels/T0002_YM_20100808140219_01.mat 114 | /trainval/images/XAM01_YM_20150727100256_01.jpg /trainval/labels/XAM01_YM_20150727100256_01.mat 115 | /trainval/images/T0001_YM_20100804160241_02.jpg /trainval/labels/T0001_YM_20100804160241_02.mat 116 | /trainval/images/T0006_XM_20120811100308_01.jpg /trainval/labels/T0006_XM_20120811100308_01.mat 117 | /trainval/images/T0001_XM_20120809100255_01.jpg /trainval/labels/T0001_XM_20120809100255_01.mat 118 | /trainval/images/T0001_XM_20130809120252_01.jpg /trainval/labels/T0001_XM_20130809120252_01.mat 119 | /trainval/images/T0001_XM_20120807160253_01.jpg /trainval/labels/T0001_XM_20120807160253_01.mat 120 | /trainval/images/T0001_XM_20120805090254_01.jpg /trainval/labels/T0001_XM_20120805090254_01.mat 121 | /trainval/images/T0001_YM_20100803150236_02.jpg /trainval/labels/T0001_YM_20100803150236_02.mat 122 | /trainval/images/T0001_YM_20100806100236_02.jpg /trainval/labels/T0001_YM_20100806100236_02.mat 123 | /trainval/images/T0006_XM_20120808100307_01.jpg /trainval/labels/T0006_XM_20120808100307_01.mat 124 | /trainval/images/T0002_YM_20100817150218_01.jpg /trainval/labels/T0002_YM_20100817150218_01.mat 125 | /trainval/images/T0001_XM_20130810100251_01.jpg /trainval/labels/T0001_XM_20130810100251_01.mat 126 | /trainval/images/T0001_XM_20120803120253_01.jpg /trainval/labels/T0001_XM_20120803120253_01.mat 127 | /trainval/images/T0001_XM_20110810130247_02.jpg /trainval/labels/T0001_XM_20110810130247_02.mat 128 | /trainval/images/T0001_XM_20110814100248_02.jpg /trainval/labels/T0001_XM_20110814100248_02.mat 129 | /trainval/images/T0001_YM_20100806140241_02.jpg /trainval/labels/T0001_YM_20100806140241_02.mat 130 | /trainval/images/T0001_XM_20120802100252_01.jpg /trainval/labels/T0001_XM_20120802100252_01.mat 131 | /trainval/images/T0001_XM_20130806120252_01.jpg /trainval/labels/T0001_XM_20130806120252_01.mat 132 | /trainval/images/T0002_YM_20100822160221_01.jpg /trainval/labels/T0002_YM_20100822160221_01.mat 133 | /trainval/images/T0001_XM_20120805140251_01.jpg /trainval/labels/T0001_XM_20120805140251_01.mat 134 | /trainval/images/T0001_XM_20120801110252_01.jpg /trainval/labels/T0001_XM_20120801110252_01.mat 135 | /trainval/images/T0001_XM_20120806090255_01.jpg /trainval/labels/T0001_XM_20120806090255_01.mat 136 | /trainval/images/T0006_XM_20120812100307_01.jpg /trainval/labels/T0006_XM_20120812100307_01.mat 137 | /trainval/images/T0002_XM_20120804160222_01.jpg /trainval/labels/T0002_XM_20120804160222_01.mat 138 | /trainval/images/T0006_XM_20120814100307_01.jpg /trainval/labels/T0006_XM_20120814100307_01.mat 139 | /trainval/images/T0002_YM_20100817130219_01.jpg /trainval/labels/T0002_YM_20100817130219_01.mat 140 | /trainval/images/T0001_YM_20100806150239_02.jpg /trainval/labels/T0001_YM_20100806150239_02.mat 141 | /trainval/images/T0001_XM_20120808160255_01.jpg /trainval/labels/T0001_XM_20120808160255_01.mat 142 | /trainval/images/T0002_XM_20120805120223_01.jpg /trainval/labels/T0002_XM_20120805120223_01.mat 143 | /trainval/images/T0006_XM_20120812130306_01.jpg /trainval/labels/T0006_XM_20120812130306_01.mat 144 | /trainval/images/T0002_XM_20120805110221_01.jpg /trainval/labels/T0002_XM_20120805110221_01.mat 145 | /trainval/images/T0002_YM_20100819160218_01.jpg /trainval/labels/T0002_YM_20100819160218_01.mat 146 | /trainval/images/T0001_XM_20120806110255_01.jpg /trainval/labels/T0001_XM_20120806110255_01.mat 147 | /trainval/images/T0001_XM_20110809100243_02.jpg /trainval/labels/T0001_XM_20110809100243_02.mat 148 | /trainval/images/T0001_YM_20100805160242_02.jpg /trainval/labels/T0001_YM_20100805160242_02.mat 149 | /trainval/images/T0001_XM_20120807100255_01.jpg /trainval/labels/T0001_XM_20120807100255_01.mat 150 | /trainval/images/T0001_XM_20120808090256_01.jpg /trainval/labels/T0001_XM_20120808090256_01.mat 151 | /trainval/images/T0002_YM_20100818160221_01.jpg /trainval/labels/T0002_YM_20100818160221_01.mat 152 | /trainval/images/XAM01_YM_20150725100254_01.jpg /trainval/labels/XAM01_YM_20150725100254_01.mat 153 | /trainval/images/T0002_YM_20100821120220_01.jpg /trainval/labels/T0002_YM_20100821120220_01.mat 154 | /trainval/images/T0001_XM_20120807140256_01.jpg /trainval/labels/T0001_XM_20120807140256_01.mat 155 | /trainval/images/T0001_XM_20120804100257_01.jpg /trainval/labels/T0001_XM_20120804100257_01.mat 156 | /trainval/images/T0001_XM_20110813160245_02.jpg /trainval/labels/T0001_XM_20110813160245_02.mat 157 | /trainval/images/T0001_XM_20110809160243_02.jpg /trainval/labels/T0001_XM_20110809160243_02.mat 158 | /trainval/images/T0001_YM_20100807160237_02.jpg /trainval/labels/T0001_YM_20100807160237_02.mat 159 | /trainval/images/T0002_XM_20120808150221_01.jpg /trainval/labels/T0002_XM_20120808150221_01.mat 160 | /trainval/images/T0002_YM_20100821160221_01.jpg /trainval/labels/T0002_YM_20100821160221_01.mat 161 | /trainval/images/T0001_YM_20100805130237_02.jpg /trainval/labels/T0001_YM_20100805130237_02.mat 162 | /trainval/images/XAM01_YM_20150730100255_01.jpg /trainval/labels/XAM01_YM_20150730100255_01.mat 163 | /trainval/images/T0001_YM_20100809140241_02.jpg /trainval/labels/T0001_YM_20100809140241_02.mat 164 | /trainval/images/T0006_XM_20120807100309_01.jpg /trainval/labels/T0006_XM_20120807100309_01.mat 165 | /trainval/images/T0001_YM_20100807110238_02.jpg /trainval/labels/T0001_YM_20100807110238_02.mat 166 | /trainval/images/T0002_XM_20120802100223_01.jpg /trainval/labels/T0002_XM_20120802100223_01.mat 167 | /trainval/images/T0006_XM_20120813160304_01.jpg /trainval/labels/T0006_XM_20120813160304_01.mat 168 | /trainval/images/T0001_XM_20120801090255_01.jpg /trainval/labels/T0001_XM_20120801090255_01.mat 169 | /trainval/images/T0001_XM_20110810160247_02.jpg /trainval/labels/T0001_XM_20110810160247_02.mat 170 | /trainval/images/T0001_XM_20110811140245_02.jpg /trainval/labels/T0001_XM_20110811140245_02.mat 171 | /trainval/images/T0001_XM_20130805130252_01.jpg /trainval/labels/T0001_XM_20130805130252_01.mat 172 | /trainval/images/T0001_XM_20130808110252_01.jpg /trainval/labels/T0001_XM_20130808110252_01.mat 173 | /trainval/images/T0001_XM_20130804080252_01.jpg /trainval/labels/T0001_XM_20130804080252_01.mat 174 | /trainval/images/T0001_XM_20120803090253_01.jpg /trainval/labels/T0001_XM_20120803090253_01.mat 175 | /trainval/images/T0001_YM_20100804110241_02.jpg /trainval/labels/T0001_YM_20100804110241_02.mat 176 | /trainval/images/XAM01_YM_20150805100255_01.jpg /trainval/labels/XAM01_YM_20150805100255_01.mat 177 | /trainval/images/T0001_XM_20120805100257_01.jpg /trainval/labels/T0001_XM_20120805100257_01.mat 178 | /trainval/images/T0001_XM_20120806150255_01.jpg /trainval/labels/T0001_XM_20120806150255_01.mat 179 | /trainval/images/T0001_XM_20110815160245_02.jpg /trainval/labels/T0001_XM_20110815160245_02.mat 180 | /trainval/images/T0001_YM_20100806110236_02.jpg /trainval/labels/T0001_YM_20100806110236_02.mat 181 | /trainval/images/T0001_XM_20120802150254_01.jpg /trainval/labels/T0001_XM_20120802150254_01.mat 182 | /trainval/images/T0001_XM_20110810100243_02.jpg /trainval/labels/T0001_XM_20110810100243_02.mat 183 | /trainval/images/T0001_XM_20120808110251_01.jpg /trainval/labels/T0001_XM_20120808110251_01.mat 184 | /trainval/images/T0001_XM_20120804160257_01.jpg /trainval/labels/T0001_XM_20120804160257_01.mat 185 | /trainval/images/T0002_XM_20120805090222_01.jpg /trainval/labels/T0002_XM_20120805090222_01.mat 186 | /trainval/images/T0002_YM_20100728120218_01.jpg /trainval/labels/T0002_YM_20100728120218_01.mat -------------------------------------------------------------------------------- /hltrainval.py: -------------------------------------------------------------------------------- 1 | """ 2 | @author: hao 3 | """ 4 | 5 | import os 6 | import argparse 7 | from time import time 8 | 9 | import cv2 as cv 10 | import numpy as np 11 | from PIL import Image 12 | from matplotlib import pyplot as plt 13 | plt.switch_backend('agg') 14 | from skimage.measure import compare_psnr 15 | from skimage.measure import compare_ssim 16 | import torch.backends.cudnn as cudnn 17 | 18 | import torch 19 | import torch.nn as nn 20 | import torch.nn.functional as F 21 | from torchvision import transforms 22 | from torch.utils.data import DataLoader 23 | from torch.optim.lr_scheduler import StepLR, MultiStepLR 24 | 25 | from hlnet import * 26 | from hldataset import * 27 | from utils import * 28 | 29 | # prevent dataloader deadlock, uncomment if deadlock occurs 30 | # cv.setNumThreads(0) 31 | cudnn.enabled = True 32 | 33 | # constant 34 | IMG_SCALE = 1./255 35 | IMG_MEAN = [.3405, .4747, .2418] 36 | IMG_STD = [1, 1, 1] 37 | SCALES = [0.7, 1, 1.3] 38 | SHORTER_SIDE = 224 39 | 40 | # system-related parameters 41 | DATA_DIR = './data/maize_counting_dataset' 42 | DATASET = 'mtc' 43 | EXP = 'tasselnetv2plus_rf110_i64o8_r0125_crop256_lr-2_bs9_epoch500' 44 | DATA_LIST = './data/maize_counting_dataset/train.txt' 45 | DATA_VAL_LIST = './data/maize_counting_dataset/test.txt' 46 | 47 | RESTORE_FROM = 'model_best.pth.tar' 48 | SNAPSHOT_DIR = './snapshots' 49 | RESULT_DIR = './results' 50 | 51 | # model-related parameters 52 | INPUT_SIZE = 64 53 | OUTPUT_STRIDE = 8 54 | MODEL = 'tasselnetv2plus' 55 | RESIZE_RATIO = 0.125 56 | 57 | # training-related parameters 58 | OPTIMIZER = 'sgd' # choice in ['sgd', 'adam'] 59 | BATCH_SIZE = 9 60 | CROP_SIZE = (256, 256) 61 | LEARNING_RATE = 1e-2 62 | MILESTONES = [200, 400] 63 | MOMENTUM = 0.95 64 | MULT = 1 65 | NUM_EPOCHS = 500 66 | NUM_CPU_WORKERS = 0 67 | PRINT_EVERY = 1 68 | RANDOM_SEED = 6 69 | WEIGHT_DECAY = 5e-4 70 | VAL_EVERY = 1 71 | 72 | device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") 73 | 74 | # add a new entry here if creating a new data loader 75 | dataset_list = { 76 | 'mtc': MaizeTasselDataset, 77 | 'wec': WhearEarDataset, 78 | 'shc': SorghumHeadDataset, 79 | } 80 | 81 | def get_arguments(): 82 | """Parse all the arguments provided from the CLI. 83 | 84 | Returns: 85 | A list of parsed arguments. 86 | """ 87 | parser = argparse.ArgumentParser(description="Object Counting Framework") 88 | # constant 89 | parser.add_argument("--image-scale", type=float, default=IMG_SCALE, help="Scale factor used in normalization.") 90 | parser.add_argument("--image-mean", nargs='+', type=float, default=IMG_MEAN, help="Mean used in normalization.") 91 | parser.add_argument("--image-std", nargs='+', type=float, default=IMG_STD, help="Std used in normalization.") 92 | parser.add_argument("--scales", type=int, default=SCALES, help="Scales of crop.") 93 | parser.add_argument("--shorter-side", type=int, default=SHORTER_SIDE, help="Shorter side of the image.") 94 | # system-related parameters 95 | parser.add_argument("--data-dir", type=str, default=DATA_DIR, help="Path to the directory containing the dataset.") 96 | parser.add_argument("--dataset", type=str, default=DATASET, help="Dataset type.") 97 | parser.add_argument("--exp", type=str, default=EXP, help="Experiment path.") 98 | parser.add_argument("--data-list", type=str, default=DATA_LIST, help="Path to the file listing the images in the dataset.") 99 | parser.add_argument("--data-val-list", type=str, default=DATA_VAL_LIST, help="Path to the file listing the images in the val dataset.") 100 | parser.add_argument("--restore-from", type=str, default=RESTORE_FROM, help="Name of restored model.") 101 | parser.add_argument("--snapshot-dir", type=str, default=SNAPSHOT_DIR, help="Where to save snapshots of the model.") 102 | parser.add_argument("--result-dir", type=str, default=RESULT_DIR, help="Where to save inferred results.") 103 | parser.add_argument("--save-output", action="store_true", help="Whether to save the output.") 104 | # model-related parameters 105 | parser.add_argument("--input-size", type=int, default=INPUT_SIZE, help="the minimum input size of the model.") 106 | parser.add_argument("--output-stride", type=int, default=OUTPUT_STRIDE, help="Output stride of the model.") 107 | parser.add_argument("--resize-ratio", type=float, default=RESIZE_RATIO, help="Resizing ratio.") 108 | parser.add_argument("--model", type=str, default=MODEL, help="model to be chosen.") 109 | parser.add_argument("--use-pretrained", action="store_true", help="Whether to use pretrained model.") 110 | parser.add_argument("--freeze-bn", action="store_true", help="Whether to freeze encoder bnorm layers.") 111 | parser.add_argument("--sync-bn", action="store_true", help="Whether to apply synchronized batch normalization.") 112 | # training-related parameters 113 | parser.add_argument("--optimizer", type=str, default=OPTIMIZER, choices=['sgd', 'adam'], help="Choose optimizer.") 114 | parser.add_argument("--batch-size", type=int, default=BATCH_SIZE, help="Number of images sent to the network in one step.") 115 | parser.add_argument("--milestones", nargs='+', type=int, default=MILESTONES, help="Multistep policy.") 116 | parser.add_argument("--crop-size", nargs='+', type=int, default=CROP_SIZE, help="Size of crop.") 117 | parser.add_argument("--evaluate-only", action="store_true", help="Whether to perform evaluation.") 118 | parser.add_argument("--learning-rate", type=float, default=LEARNING_RATE, help="Base learning rate for training.") 119 | parser.add_argument("--momentum", type=float, default=MOMENTUM, help="Momentum component of the optimizer.") 120 | parser.add_argument("--weight-decay", type=float, default=WEIGHT_DECAY, help="Regularisation parameter for L2-loss.") 121 | parser.add_argument("--mult", type=float, default=MULT, help="LR multiplier for pretrained layers.") 122 | parser.add_argument("--num-epochs", type=int, default=NUM_EPOCHS, help="Number of training epochs.") 123 | parser.add_argument("--num-workers", type=int, default=NUM_CPU_WORKERS, help="Number of CPU cores used.") 124 | parser.add_argument("--print-every", type=int, default=PRINT_EVERY, help="Print information every often.") 125 | parser.add_argument("--random-seed", type=int, default=RANDOM_SEED, help="Random seed to have reproducible results.") 126 | parser.add_argument("--val-every", type=int, default=VAL_EVERY, help="How often performing validation.") 127 | return parser.parse_args() 128 | 129 | def save_checkpoint(state, snapshot_dir, filename='model_ckpt.pth.tar'): 130 | torch.save(state, '{}/{}'.format(snapshot_dir, filename)) 131 | 132 | def plot_learning_curves(net, dir_to_save): 133 | # plot learning curves 134 | fig = plt.figure(figsize=(16, 9)) 135 | ax1 = fig.add_subplot(1, 2, 1) 136 | ax1.plot(net.train_loss['epoch_loss'], label='train loss', color='tab:blue') 137 | ax1.legend(loc = 'upper right') 138 | ax2 = fig.add_subplot(1, 2, 2) 139 | ax2.plot(net.val_loss['epoch_loss'], label='val mae', color='tab:orange') 140 | ax2.legend(loc = 'upper right') 141 | # ax2.set_ylim((0,50)) 142 | fig.savefig(os.path.join(dir_to_save, 'learning_curves.png'), bbox_inches='tight', dpi = 300) 143 | plt.close() 144 | 145 | def train(net, train_loader, criterion, optimizer, epoch, args): 146 | # switch to 'train' mode 147 | net.train() 148 | 149 | # uncomment the following line if the training images don't have the same size 150 | cudnn.benchmark = True 151 | 152 | if args.batch_size == 1: 153 | for m in net.modules(): 154 | if isinstance(m, nn.BatchNorm2d): 155 | m.eval() 156 | 157 | running_loss = 0.0 158 | avg_frame_rate = 0.0 159 | in_sz = args.input_size 160 | os = args.output_stride 161 | target_filter = torch.cuda.FloatTensor(1, 1, in_sz, in_sz).fill_(1) 162 | for i, sample in enumerate(train_loader): 163 | torch.cuda.synchronize() 164 | start = time() 165 | 166 | inputs, targets = sample['image'], sample['target'] 167 | inputs, targets = inputs.cuda(), targets.cuda() 168 | 169 | # zero the parameter gradients 170 | optimizer.zero_grad() 171 | 172 | # forward 173 | outputs = net(inputs, is_normalize=False) 174 | # generate targets 175 | targets = F.conv2d(targets, target_filter, stride=os) 176 | # compute loss 177 | loss = criterion(outputs, targets) 178 | 179 | # backward + optimize 180 | loss.backward() 181 | optimizer.step() 182 | # collect and print statistics 183 | running_loss += loss.item() 184 | 185 | torch.cuda.synchronize() 186 | end = time() 187 | 188 | running_frame_rate = args.batch_size * float(1 / (end - start)) 189 | avg_frame_rate = (avg_frame_rate*i + running_frame_rate)/(i+1) 190 | if i % args.print_every == args.print_every-1: 191 | print('epoch: %d, train: %d/%d, ' 192 | 'loss: %.5f, frame: %.2fHz/%.2fHz' % ( 193 | epoch, 194 | i+1, 195 | len(train_loader), 196 | running_loss / (i+1), 197 | running_frame_rate, 198 | avg_frame_rate 199 | )) 200 | net.train_loss['epoch_loss'].append(running_loss / (i+1)) 201 | 202 | 203 | def validate(net, valset, val_loader, criterion, epoch, args): 204 | # switch to 'eval' mode 205 | net.eval() 206 | cudnn.benchmark = False 207 | 208 | image_list = valset.image_list 209 | 210 | if args.save_output: 211 | epoch_result_dir = os.path.join(args.result_dir, str(epoch)) 212 | if not os.path.exists(epoch_result_dir): 213 | os.makedirs(epoch_result_dir) 214 | cmap = plt.cm.get_cmap('jet') 215 | 216 | pd_counts = [] 217 | gt_counts = [] 218 | with torch.no_grad(): 219 | avg_frame_rate = 0.0 220 | for i, sample in enumerate(val_loader): 221 | torch.cuda.synchronize() 222 | start = time() 223 | 224 | image, gtcount = sample['image'], sample['gtcount'] 225 | # inference 226 | output = net(image.cuda(), is_normalize=not args.save_output) 227 | if args.save_output: 228 | output_save = output 229 | # normalization 230 | output = Normalizer.gpu_normalizer(output, image.size()[2], image.size()[3], args.input_size, args.output_stride) 231 | # postprocessing 232 | output = np.clip(output, 0, None) 233 | 234 | pdcount = output.sum() 235 | gtcount = float(gtcount.numpy()) 236 | 237 | if args.save_output: 238 | _, image_name = os.path.split(image_list[i]) 239 | output_save = np.clip(output_save.squeeze().cpu().numpy(), 0, None) 240 | output_save = recover_countmap(output_save, image, args.input_size, args.output_stride) 241 | output_save = output_save / (output_save.max() + 1e-12) 242 | output_save = cmap(output_save) * 255. 243 | # image composition 244 | image = valset.images[image_list[i]] 245 | nh, nw = output_save.shape[:2] 246 | image = cv2.resize(image, (nw, nh), interpolation = cv2.INTER_CUBIC) 247 | output_save = 0.5 * image + 0.5 * output_save[:, :, 0:3] 248 | 249 | dotimage = valset.dotimages[image_list[i]] 250 | 251 | fig = plt.figure() 252 | ax1 = fig.add_subplot(1, 2, 1) 253 | ax1.imshow(dotimage.astype(np.uint8)) 254 | ax1.get_xaxis().set_visible(False) 255 | ax1.get_yaxis().set_visible(False) 256 | ax2 = fig.add_subplot(1, 2, 2) 257 | ax2.imshow(output_save.astype(np.uint8)) 258 | ax2.get_xaxis().set_visible(False) 259 | ax2.get_yaxis().set_visible(False) 260 | fig.suptitle('manual count=%4.2f, inferred count=%4.2f'%(gtcount, pdcount), fontsize=10) 261 | if args.dataset == 'mtc': 262 | plt.tight_layout(rect=[0, 0, 1, 1.4]) # maize tassels counting 263 | elif args.dataset == 'wec': 264 | plt.tight_layout(rect=[0, 0, 1, 1.45]) # wheat ears counting 265 | elif args.dataset == 'shc': 266 | plt.tight_layout(rect=[0, 0, 0.95, 1]) # sorghum heads counting -- dataset1 267 | # plt.tight_layout(rect=[0, 0, 1.2, 1]) # sorghum heads counting -- dataset2 268 | plt.savefig(os.path.join(epoch_result_dir, image_name.replace('.jpg', '.png')), bbox_inches='tight', dpi = 300) 269 | plt.close() 270 | 271 | # compute mae and mse 272 | pd_counts.append(pdcount) 273 | gt_counts.append(gtcount) 274 | mae = compute_mae(pd_counts, gt_counts) 275 | mse = compute_mse(pd_counts, gt_counts) 276 | rmae, rmse = compute_relerr(pd_counts, gt_counts) 277 | 278 | torch.cuda.synchronize() 279 | end = time() 280 | 281 | running_frame_rate = 1 * float(1 / (end - start)) 282 | avg_frame_rate = (avg_frame_rate*i + running_frame_rate)/(i+1) 283 | if i % args.print_every == args.print_every - 1: 284 | print( 285 | 'epoch: {0}, test: {1}/{2}, pre: {3:.2f}, gt:{4:.2f}, me:{5:.2f}, mae: {6:.2f}, mse: {7:.2f}, rmae: {8:.2f}%, rmse: {9:.2f}%, frame: {10:.2f}Hz/{11:.2f}Hz' 286 | .format(epoch, i+1, len(val_loader), pdcount, gtcount, pdcount-gtcount, mae, mse, rmae, rmse, running_frame_rate, avg_frame_rate) 287 | ) 288 | start = time() 289 | r2 = rsquared(pd_counts, gt_counts) 290 | np.save(args.snapshot_dir+'/pd.npy', pd_counts) 291 | np.save(args.snapshot_dir+'/gt.npy', gt_counts) 292 | print('epoch: {0}, mae: {1:.2f}, mse: {2:.2f}, rmae: {3:.2f}%, rmse: {4:.2f}%, r2: {5:.4f}'.format(epoch, mae, mse, rmae, rmse, r2)) 293 | # write to files 294 | with open(os.path.join(args.snapshot_dir, args.exp+'.txt'), 'a') as f: 295 | print( 296 | 'epoch: {0}, mae: {1:.2f}, mse: {2:.2f}, rmae: {3:.2f}%, rmse: {4:.2f}%, r2: {5:.4f}'.format(epoch, mae, mse, rmae, rmse, r2), 297 | file=f 298 | ) 299 | with open(os.path.join(args.snapshot_dir, 'counts.txt'), 'a') as f: 300 | for pd, gt in zip(pd_counts, gt_counts): 301 | print( 302 | '{0} {1}'.format(pd, gt), 303 | file=f 304 | ) 305 | # save stats 306 | net.val_loss['epoch_loss'].append(mae) 307 | net.measure['mae'].append(mae) 308 | net.measure['mse'].append(mse) 309 | net.measure['rmae'].append(rmae) 310 | net.measure['rmse'].append(rmse) 311 | net.measure['r2'].append(r2) 312 | 313 | def main(): 314 | args = get_arguments() 315 | 316 | # args.evaluate_only = True 317 | # args.save_output = True 318 | 319 | args.image_mean = np.array(args.image_mean).reshape((1, 1, 3)) 320 | args.image_std = np.array(args.image_std).reshape((1, 1, 3)) 321 | 322 | args.crop_size = tuple(args.crop_size) if len(args.crop_size) > 1 else args.crop_size 323 | 324 | # seeding for reproducbility 325 | if torch.cuda.is_available(): 326 | torch.cuda.manual_seed(args.random_seed) 327 | torch.manual_seed(args.random_seed) 328 | np.random.seed(args.random_seed) 329 | 330 | # instantiate dataset 331 | dataset = dataset_list[args.dataset] 332 | 333 | args.snapshot_dir = os.path.join(args.snapshot_dir, args.dataset.lower(), args.exp) 334 | if not os.path.exists(args.snapshot_dir): 335 | os.makedirs(args.snapshot_dir) 336 | 337 | args.result_dir = os.path.join(args.result_dir, args.dataset.lower(), args.exp) 338 | if not os.path.exists(args.result_dir): 339 | os.makedirs(args.result_dir) 340 | 341 | args.restore_from = os.path.join(args.snapshot_dir, args.restore_from) 342 | 343 | arguments = vars(args) 344 | for item in arguments: 345 | print(item, ':\t' , arguments[item]) 346 | 347 | # instantiate network 348 | net = CountingModels( 349 | arc=args.model, 350 | input_size=args.input_size, 351 | output_stride=args.output_stride 352 | ) 353 | 354 | net = nn.DataParallel(net) 355 | net.cuda() 356 | 357 | # filter parameters 358 | learning_params = [p[1] for p in net.named_parameters()] 359 | pretrained_params = [] 360 | 361 | # define loss function and optimizer 362 | criterion = nn.L1Loss(reduction='mean').cuda() 363 | 364 | if args.optimizer == 'sgd': 365 | optimizer = torch.optim.SGD( 366 | [ 367 | {'params': learning_params}, 368 | {'params': pretrained_params, 'lr': args.learning_rate / args.mult}, 369 | ], 370 | lr=args.learning_rate, 371 | momentum=args.momentum, 372 | weight_decay=args.weight_decay 373 | ) 374 | elif args.optimizer == 'adam': 375 | optimizer = torch.optim.Adam( 376 | [ 377 | {'params': learning_params}, 378 | {'params': pretrained_params, 'lr': args.learning_rate / args.mult}, 379 | ], 380 | lr=args.learning_rate 381 | ) 382 | else: 383 | raise NotImplementedError 384 | 385 | # restore parameters 386 | start_epoch = 0 387 | net.train_loss = { 388 | 'running_loss': [], 389 | 'epoch_loss': [] 390 | } 391 | net.val_loss = { 392 | 'running_loss': [], 393 | 'epoch_loss': [] 394 | } 395 | net.measure = { 396 | 'mae': [], 397 | 'mse': [], 398 | 'rmae': [], 399 | 'rmse': [], 400 | 'r2': [] 401 | } 402 | if args.restore_from is not None: 403 | if os.path.isfile(args.restore_from): 404 | checkpoint = torch.load(args.restore_from) 405 | net.load_state_dict(checkpoint['state_dict']) 406 | if 'epoch' in checkpoint: 407 | start_epoch = checkpoint['epoch'] 408 | if 'optimizer' in checkpoint: 409 | optimizer.load_state_dict(checkpoint['optimizer']) 410 | if 'train_loss' in checkpoint: 411 | net.train_loss = checkpoint['train_loss'] 412 | if 'val_loss' in checkpoint: 413 | net.val_loss = checkpoint['val_loss'] 414 | if 'measure' in checkpoint: 415 | net.measure['mae'] = checkpoint['measure']['mae'] if 'mae' in checkpoint['measure'] else [] 416 | net.measure['mse'] = checkpoint['measure']['mse'] if 'mse' in checkpoint['measure'] else [] 417 | net.measure['rmae'] = checkpoint['measure']['rmae'] if 'rmae' in checkpoint['measure'] else [] 418 | net.measure['rmse'] = checkpoint['measure']['rmse'] if 'rmse' in checkpoint['measure'] else [] 419 | net.measure['r2'] = checkpoint['measure']['r2'] if 'r2' in checkpoint['measure'] else [] 420 | print("==> load checkpoint '{}' (epoch {})" 421 | .format(args.restore_from, start_epoch)) 422 | else: 423 | with open(os.path.join(args.snapshot_dir, args.exp+'.txt'), 'a') as f: 424 | for item in arguments: 425 | print(item, ':\t' , arguments[item], file=f) 426 | print("==> no checkpoint found at '{}'".format(args.restore_from)) 427 | 428 | # define transform 429 | transform_train = [ 430 | RandomCrop(args.crop_size), 431 | RandomFlip(), 432 | Normalize( 433 | args.image_scale, 434 | args.image_mean, 435 | args.image_std 436 | ), 437 | ToTensor(), 438 | ZeroPadding(args.output_stride) 439 | ] 440 | transform_val = [ 441 | Normalize( 442 | args.image_scale, 443 | args.image_mean, 444 | args.image_std 445 | ), 446 | ToTensor(), 447 | ZeroPadding(args.output_stride) 448 | ] 449 | composed_transform_train = transforms.Compose(transform_train) 450 | composed_transform_val = transforms.Compose(transform_val) 451 | 452 | # define dataset loader 453 | trainset = dataset( 454 | data_dir=args.data_dir, 455 | data_list=args.data_list, 456 | ratio=args.resize_ratio, 457 | train=True, 458 | transform=composed_transform_train 459 | ) 460 | train_loader = DataLoader( 461 | trainset, 462 | batch_size=args.batch_size, 463 | shuffle=True, 464 | num_workers=args.num_workers, 465 | pin_memory=True, 466 | drop_last=True 467 | ) 468 | valset = dataset( 469 | data_dir=args.data_dir, 470 | data_list=args.data_val_list, 471 | ratio=args.resize_ratio, 472 | train=False, 473 | transform=composed_transform_val 474 | ) 475 | val_loader = DataLoader( 476 | valset, 477 | batch_size=1, 478 | shuffle=False, 479 | num_workers=args.num_workers, 480 | pin_memory=True 481 | ) 482 | 483 | print('alchemy start...') 484 | if args.evaluate_only: 485 | validate(net, valset, val_loader, criterion, start_epoch, args) 486 | return 487 | 488 | best_mae = 1000000.0 489 | resume_epoch = -1 if start_epoch == 0 else start_epoch 490 | scheduler = MultiStepLR(optimizer, milestones=args.milestones, gamma=0.1, last_epoch=resume_epoch) 491 | for epoch in range(start_epoch, args.num_epochs): 492 | # train 493 | train(net, train_loader, criterion, optimizer, epoch+1, args) 494 | if epoch % args.val_every == args.val_every - 1: 495 | # val 496 | validate(net, valset, val_loader, criterion, epoch+1, args) 497 | # save_checkpoint 498 | state = { 499 | 'state_dict': net.state_dict(), 500 | 'optimizer': optimizer.state_dict(), 501 | 'epoch': epoch+1, 502 | 'train_loss': net.train_loss, 503 | 'val_loss': net.val_loss, 504 | 'measure': net.measure 505 | } 506 | save_checkpoint(state, args.snapshot_dir, filename='model_ckpt.pth.tar') 507 | if net.measure['mae'][-1] <= best_mae: 508 | save_checkpoint(state, args.snapshot_dir, filename='model_best.pth.tar') 509 | best_mae = net.measure['mae'][-1] 510 | best_mse = net.measure['mse'][-1] 511 | best_rmae = net.measure['rmae'][-1] 512 | best_rmse = net.measure['rmse'][-1] 513 | best_r2 = net.measure['r2'][-1] 514 | print(args.exp+' epoch {} finished!'.format(epoch+1)) 515 | print('best mae: {0:.2f}, best mse: {1:.2f}, best_rmae: {2:.2f}, best_rmse: {3:.2f}, best_r2: {4:.4f}' 516 | .format(best_mae, best_mse, best_rmae, best_rmse, best_r2)) 517 | plot_learning_curves(net, args.snapshot_dir) 518 | scheduler.step() 519 | 520 | print('Experiments with '+args.exp+' done!') 521 | with open(os.path.join(args.snapshot_dir, args.exp+'.txt'), 'a') as f: 522 | print( 523 | 'best mae: {0:.2f}, best mse: {1:.2f}, best_rmae: {2:.2f}, best_rmse: {3:.2f}, best_r2: {4:.4f}' 524 | .format(best_mae, best_mse, best_rmae, best_rmse, best_r2), 525 | file=f 526 | ) 527 | print( 528 | 'overall best mae: {0:.2f}, overall best mse: {1:.2f}, overall best_rmae: {2:.2f}, overall best_rmse: {3:.2f}, overall best_r2: {4:.4f}' 529 | .format(min(net.measure['mae']), min(net.measure['mse']), min(net.measure['rmae']), min(net.measure['rmse']), max(net.measure['r2'])), 530 | file=f 531 | ) 532 | 533 | if __name__ == "__main__": 534 | main() -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | GNU GENERAL PUBLIC LICENSE 2 | Version 3, 29 June 2007 3 | 4 | Copyright (C) 2007 Free Software Foundation, Inc. 5 | Everyone is permitted to copy and distribute verbatim copies 6 | of this license document, but changing it is not allowed. 7 | 8 | Preamble 9 | 10 | The GNU General Public License is a free, copyleft license for 11 | software and other kinds of works. 12 | 13 | The licenses for most software and other practical works are designed 14 | to take away your freedom to share and change the works. By contrast, 15 | the GNU General Public License is intended to guarantee your freedom to 16 | share and change all versions of a program--to make sure it remains free 17 | software for all its users. We, the Free Software Foundation, use the 18 | GNU General Public License for most of our software; it applies also to 19 | any other work released this way by its authors. You can apply it to 20 | your programs, too. 21 | 22 | When we speak of free software, we are referring to freedom, not 23 | price. Our General Public Licenses are designed to make sure that you 24 | have the freedom to distribute copies of free software (and charge for 25 | them if you wish), that you receive source code or can get it if you 26 | want it, that you can change the software or use pieces of it in new 27 | free programs, and that you know you can do these things. 28 | 29 | To protect your rights, we need to prevent others from denying you 30 | these rights or asking you to surrender the rights. Therefore, you have 31 | certain responsibilities if you distribute copies of the software, or if 32 | you modify it: responsibilities to respect the freedom of others. 33 | 34 | For example, if you distribute copies of such a program, whether 35 | gratis or for a fee, you must pass on to the recipients the same 36 | freedoms that you received. You must make sure that they, too, receive 37 | or can get the source code. And you must show them these terms so they 38 | know their rights. 39 | 40 | Developers that use the GNU GPL protect your rights with two steps: 41 | (1) assert copyright on the software, and (2) offer you this License 42 | giving you legal permission to copy, distribute and/or modify it. 43 | 44 | For the developers' and authors' protection, the GPL clearly explains 45 | that there is no warranty for this free software. For both users' and 46 | authors' sake, the GPL requires that modified versions be marked as 47 | changed, so that their problems will not be attributed erroneously to 48 | authors of previous versions. 49 | 50 | Some devices are designed to deny users access to install or run 51 | modified versions of the software inside them, although the manufacturer 52 | can do so. This is fundamentally incompatible with the aim of 53 | protecting users' freedom to change the software. The systematic 54 | pattern of such abuse occurs in the area of products for individuals to 55 | use, which is precisely where it is most unacceptable. Therefore, we 56 | have designed this version of the GPL to prohibit the practice for those 57 | products. If such problems arise substantially in other domains, we 58 | stand ready to extend this provision to those domains in future versions 59 | of the GPL, as needed to protect the freedom of users. 60 | 61 | Finally, every program is threatened constantly by software patents. 62 | States should not allow patents to restrict development and use of 63 | software on general-purpose computers, but in those that do, we wish to 64 | avoid the special danger that patents applied to a free program could 65 | make it effectively proprietary. To prevent this, the GPL assures that 66 | patents cannot be used to render the program non-free. 67 | 68 | The precise terms and conditions for copying, distribution and 69 | modification follow. 70 | 71 | TERMS AND CONDITIONS 72 | 73 | 0. Definitions. 74 | 75 | "This License" refers to version 3 of the GNU General Public License. 76 | 77 | "Copyright" also means copyright-like laws that apply to other kinds of 78 | works, such as semiconductor masks. 79 | 80 | "The Program" refers to any copyrightable work licensed under this 81 | License. Each licensee is addressed as "you". "Licensees" and 82 | "recipients" may be individuals or organizations. 83 | 84 | To "modify" a work means to copy from or adapt all or part of the work 85 | in a fashion requiring copyright permission, other than the making of an 86 | exact copy. The resulting work is called a "modified version" of the 87 | earlier work or a work "based on" the earlier work. 88 | 89 | A "covered work" means either the unmodified Program or a work based 90 | on the Program. 91 | 92 | To "propagate" a work means to do anything with it that, without 93 | permission, would make you directly or secondarily liable for 94 | infringement under applicable copyright law, except executing it on a 95 | computer or modifying a private copy. Propagation includes copying, 96 | distribution (with or without modification), making available to the 97 | public, and in some countries other activities as well. 98 | 99 | To "convey" a work means any kind of propagation that enables other 100 | parties to make or receive copies. Mere interaction with a user through 101 | a computer network, with no transfer of a copy, is not conveying. 102 | 103 | An interactive user interface displays "Appropriate Legal Notices" 104 | to the extent that it includes a convenient and prominently visible 105 | feature that (1) displays an appropriate copyright notice, and (2) 106 | tells the user that there is no warranty for the work (except to the 107 | extent that warranties are provided), that licensees may convey the 108 | work under this License, and how to view a copy of this License. If 109 | the interface presents a list of user commands or options, such as a 110 | menu, a prominent item in the list meets this criterion. 111 | 112 | 1. Source Code. 113 | 114 | The "source code" for a work means the preferred form of the work 115 | for making modifications to it. "Object code" means any non-source 116 | form of a work. 117 | 118 | A "Standard Interface" means an interface that either is an official 119 | standard defined by a recognized standards body, or, in the case of 120 | interfaces specified for a particular programming language, one that 121 | is widely used among developers working in that language. 122 | 123 | The "System Libraries" of an executable work include anything, other 124 | than the work as a whole, that (a) is included in the normal form of 125 | packaging a Major Component, but which is not part of that Major 126 | Component, and (b) serves only to enable use of the work with that 127 | Major Component, or to implement a Standard Interface for which an 128 | implementation is available to the public in source code form. A 129 | "Major Component", in this context, means a major essential component 130 | (kernel, window system, and so on) of the specific operating system 131 | (if any) on which the executable work runs, or a compiler used to 132 | produce the work, or an object code interpreter used to run it. 133 | 134 | The "Corresponding Source" for a work in object code form means all 135 | the source code needed to generate, install, and (for an executable 136 | work) run the object code and to modify the work, including scripts to 137 | control those activities. However, it does not include the work's 138 | System Libraries, or general-purpose tools or generally available free 139 | programs which are used unmodified in performing those activities but 140 | which are not part of the work. For example, Corresponding Source 141 | includes interface definition files associated with source files for 142 | the work, and the source code for shared libraries and dynamically 143 | linked subprograms that the work is specifically designed to require, 144 | such as by intimate data communication or control flow between those 145 | subprograms and other parts of the work. 146 | 147 | The Corresponding Source need not include anything that users 148 | can regenerate automatically from other parts of the Corresponding 149 | Source. 150 | 151 | The Corresponding Source for a work in source code form is that 152 | same work. 153 | 154 | 2. Basic Permissions. 155 | 156 | All rights granted under this License are granted for the term of 157 | copyright on the Program, and are irrevocable provided the stated 158 | conditions are met. This License explicitly affirms your unlimited 159 | permission to run the unmodified Program. The output from running a 160 | covered work is covered by this License only if the output, given its 161 | content, constitutes a covered work. This License acknowledges your 162 | rights of fair use or other equivalent, as provided by copyright law. 163 | 164 | You may make, run and propagate covered works that you do not 165 | convey, without conditions so long as your license otherwise remains 166 | in force. You may convey covered works to others for the sole purpose 167 | of having them make modifications exclusively for you, or provide you 168 | with facilities for running those works, provided that you comply with 169 | the terms of this License in conveying all material for which you do 170 | not control copyright. Those thus making or running the covered works 171 | for you must do so exclusively on your behalf, under your direction 172 | and control, on terms that prohibit them from making any copies of 173 | your copyrighted material outside their relationship with you. 174 | 175 | Conveying under any other circumstances is permitted solely under 176 | the conditions stated below. Sublicensing is not allowed; section 10 177 | makes it unnecessary. 178 | 179 | 3. Protecting Users' Legal Rights From Anti-Circumvention Law. 180 | 181 | No covered work shall be deemed part of an effective technological 182 | measure under any applicable law fulfilling obligations under article 183 | 11 of the WIPO copyright treaty adopted on 20 December 1996, or 184 | similar laws prohibiting or restricting circumvention of such 185 | measures. 186 | 187 | When you convey a covered work, you waive any legal power to forbid 188 | circumvention of technological measures to the extent such circumvention 189 | is effected by exercising rights under this License with respect to 190 | the covered work, and you disclaim any intention to limit operation or 191 | modification of the work as a means of enforcing, against the work's 192 | users, your or third parties' legal rights to forbid circumvention of 193 | technological measures. 194 | 195 | 4. Conveying Verbatim Copies. 196 | 197 | You may convey verbatim copies of the Program's source code as you 198 | receive it, in any medium, provided that you conspicuously and 199 | appropriately publish on each copy an appropriate copyright notice; 200 | keep intact all notices stating that this License and any 201 | non-permissive terms added in accord with section 7 apply to the code; 202 | keep intact all notices of the absence of any warranty; and give all 203 | recipients a copy of this License along with the Program. 204 | 205 | You may charge any price or no price for each copy that you convey, 206 | and you may offer support or warranty protection for a fee. 207 | 208 | 5. Conveying Modified Source Versions. 209 | 210 | You may convey a work based on the Program, or the modifications to 211 | produce it from the Program, in the form of source code under the 212 | terms of section 4, provided that you also meet all of these conditions: 213 | 214 | a) The work must carry prominent notices stating that you modified 215 | it, and giving a relevant date. 216 | 217 | b) The work must carry prominent notices stating that it is 218 | released under this License and any conditions added under section 219 | 7. This requirement modifies the requirement in section 4 to 220 | "keep intact all notices". 221 | 222 | c) You must license the entire work, as a whole, under this 223 | License to anyone who comes into possession of a copy. This 224 | License will therefore apply, along with any applicable section 7 225 | additional terms, to the whole of the work, and all its parts, 226 | regardless of how they are packaged. This License gives no 227 | permission to license the work in any other way, but it does not 228 | invalidate such permission if you have separately received it. 229 | 230 | d) If the work has interactive user interfaces, each must display 231 | Appropriate Legal Notices; however, if the Program has interactive 232 | interfaces that do not display Appropriate Legal Notices, your 233 | work need not make them do so. 234 | 235 | A compilation of a covered work with other separate and independent 236 | works, which are not by their nature extensions of the covered work, 237 | and which are not combined with it such as to form a larger program, 238 | in or on a volume of a storage or distribution medium, is called an 239 | "aggregate" if the compilation and its resulting copyright are not 240 | used to limit the access or legal rights of the compilation's users 241 | beyond what the individual works permit. Inclusion of a covered work 242 | in an aggregate does not cause this License to apply to the other 243 | parts of the aggregate. 244 | 245 | 6. Conveying Non-Source Forms. 246 | 247 | You may convey a covered work in object code form under the terms 248 | of sections 4 and 5, provided that you also convey the 249 | machine-readable Corresponding Source under the terms of this License, 250 | in one of these ways: 251 | 252 | a) Convey the object code in, or embodied in, a physical product 253 | (including a physical distribution medium), accompanied by the 254 | Corresponding Source fixed on a durable physical medium 255 | customarily used for software interchange. 256 | 257 | b) Convey the object code in, or embodied in, a physical product 258 | (including a physical distribution medium), accompanied by a 259 | written offer, valid for at least three years and valid for as 260 | long as you offer spare parts or customer support for that product 261 | model, to give anyone who possesses the object code either (1) a 262 | copy of the Corresponding Source for all the software in the 263 | product that is covered by this License, on a durable physical 264 | medium customarily used for software interchange, for a price no 265 | more than your reasonable cost of physically performing this 266 | conveying of source, or (2) access to copy the 267 | Corresponding Source from a network server at no charge. 268 | 269 | c) Convey individual copies of the object code with a copy of the 270 | written offer to provide the Corresponding Source. This 271 | alternative is allowed only occasionally and noncommercially, and 272 | only if you received the object code with such an offer, in accord 273 | with subsection 6b. 274 | 275 | d) Convey the object code by offering access from a designated 276 | place (gratis or for a charge), and offer equivalent access to the 277 | Corresponding Source in the same way through the same place at no 278 | further charge. You need not require recipients to copy the 279 | Corresponding Source along with the object code. If the place to 280 | copy the object code is a network server, the Corresponding Source 281 | may be on a different server (operated by you or a third party) 282 | that supports equivalent copying facilities, provided you maintain 283 | clear directions next to the object code saying where to find the 284 | Corresponding Source. Regardless of what server hosts the 285 | Corresponding Source, you remain obligated to ensure that it is 286 | available for as long as needed to satisfy these requirements. 287 | 288 | e) Convey the object code using peer-to-peer transmission, provided 289 | you inform other peers where the object code and Corresponding 290 | Source of the work are being offered to the general public at no 291 | charge under subsection 6d. 292 | 293 | A separable portion of the object code, whose source code is excluded 294 | from the Corresponding Source as a System Library, need not be 295 | included in conveying the object code work. 296 | 297 | A "User Product" is either (1) a "consumer product", which means any 298 | tangible personal property which is normally used for personal, family, 299 | or household purposes, or (2) anything designed or sold for incorporation 300 | into a dwelling. In determining whether a product is a consumer product, 301 | doubtful cases shall be resolved in favor of coverage. For a particular 302 | product received by a particular user, "normally used" refers to a 303 | typical or common use of that class of product, regardless of the status 304 | of the particular user or of the way in which the particular user 305 | actually uses, or expects or is expected to use, the product. A product 306 | is a consumer product regardless of whether the product has substantial 307 | commercial, industrial or non-consumer uses, unless such uses represent 308 | the only significant mode of use of the product. 309 | 310 | "Installation Information" for a User Product means any methods, 311 | procedures, authorization keys, or other information required to install 312 | and execute modified versions of a covered work in that User Product from 313 | a modified version of its Corresponding Source. The information must 314 | suffice to ensure that the continued functioning of the modified object 315 | code is in no case prevented or interfered with solely because 316 | modification has been made. 317 | 318 | If you convey an object code work under this section in, or with, or 319 | specifically for use in, a User Product, and the conveying occurs as 320 | part of a transaction in which the right of possession and use of the 321 | User Product is transferred to the recipient in perpetuity or for a 322 | fixed term (regardless of how the transaction is characterized), the 323 | Corresponding Source conveyed under this section must be accompanied 324 | by the Installation Information. But this requirement does not apply 325 | if neither you nor any third party retains the ability to install 326 | modified object code on the User Product (for example, the work has 327 | been installed in ROM). 328 | 329 | The requirement to provide Installation Information does not include a 330 | requirement to continue to provide support service, warranty, or updates 331 | for a work that has been modified or installed by the recipient, or for 332 | the User Product in which it has been modified or installed. Access to a 333 | network may be denied when the modification itself materially and 334 | adversely affects the operation of the network or violates the rules and 335 | protocols for communication across the network. 336 | 337 | Corresponding Source conveyed, and Installation Information provided, 338 | in accord with this section must be in a format that is publicly 339 | documented (and with an implementation available to the public in 340 | source code form), and must require no special password or key for 341 | unpacking, reading or copying. 342 | 343 | 7. Additional Terms. 344 | 345 | "Additional permissions" are terms that supplement the terms of this 346 | License by making exceptions from one or more of its conditions. 347 | Additional permissions that are applicable to the entire Program shall 348 | be treated as though they were included in this License, to the extent 349 | that they are valid under applicable law. If additional permissions 350 | apply only to part of the Program, that part may be used separately 351 | under those permissions, but the entire Program remains governed by 352 | this License without regard to the additional permissions. 353 | 354 | When you convey a copy of a covered work, you may at your option 355 | remove any additional permissions from that copy, or from any part of 356 | it. (Additional permissions may be written to require their own 357 | removal in certain cases when you modify the work.) You may place 358 | additional permissions on material, added by you to a covered work, 359 | for which you have or can give appropriate copyright permission. 360 | 361 | Notwithstanding any other provision of this License, for material you 362 | add to a covered work, you may (if authorized by the copyright holders of 363 | that material) supplement the terms of this License with terms: 364 | 365 | a) Disclaiming warranty or limiting liability differently from the 366 | terms of sections 15 and 16 of this License; or 367 | 368 | b) Requiring preservation of specified reasonable legal notices or 369 | author attributions in that material or in the Appropriate Legal 370 | Notices displayed by works containing it; or 371 | 372 | c) Prohibiting misrepresentation of the origin of that material, or 373 | requiring that modified versions of such material be marked in 374 | reasonable ways as different from the original version; or 375 | 376 | d) Limiting the use for publicity purposes of names of licensors or 377 | authors of the material; or 378 | 379 | e) Declining to grant rights under trademark law for use of some 380 | trade names, trademarks, or service marks; or 381 | 382 | f) Requiring indemnification of licensors and authors of that 383 | material by anyone who conveys the material (or modified versions of 384 | it) with contractual assumptions of liability to the recipient, for 385 | any liability that these contractual assumptions directly impose on 386 | those licensors and authors. 387 | 388 | All other non-permissive additional terms are considered "further 389 | restrictions" within the meaning of section 10. If the Program as you 390 | received it, or any part of it, contains a notice stating that it is 391 | governed by this License along with a term that is a further 392 | restriction, you may remove that term. If a license document contains 393 | a further restriction but permits relicensing or conveying under this 394 | License, you may add to a covered work material governed by the terms 395 | of that license document, provided that the further restriction does 396 | not survive such relicensing or conveying. 397 | 398 | If you add terms to a covered work in accord with this section, you 399 | must place, in the relevant source files, a statement of the 400 | additional terms that apply to those files, or a notice indicating 401 | where to find the applicable terms. 402 | 403 | Additional terms, permissive or non-permissive, may be stated in the 404 | form of a separately written license, or stated as exceptions; 405 | the above requirements apply either way. 406 | 407 | 8. Termination. 408 | 409 | You may not propagate or modify a covered work except as expressly 410 | provided under this License. Any attempt otherwise to propagate or 411 | modify it is void, and will automatically terminate your rights under 412 | this License (including any patent licenses granted under the third 413 | paragraph of section 11). 414 | 415 | However, if you cease all violation of this License, then your 416 | license from a particular copyright holder is reinstated (a) 417 | provisionally, unless and until the copyright holder explicitly and 418 | finally terminates your license, and (b) permanently, if the copyright 419 | holder fails to notify you of the violation by some reasonable means 420 | prior to 60 days after the cessation. 421 | 422 | Moreover, your license from a particular copyright holder is 423 | reinstated permanently if the copyright holder notifies you of the 424 | violation by some reasonable means, this is the first time you have 425 | received notice of violation of this License (for any work) from that 426 | copyright holder, and you cure the violation prior to 30 days after 427 | your receipt of the notice. 428 | 429 | Termination of your rights under this section does not terminate the 430 | licenses of parties who have received copies or rights from you under 431 | this License. If your rights have been terminated and not permanently 432 | reinstated, you do not qualify to receive new licenses for the same 433 | material under section 10. 434 | 435 | 9. Acceptance Not Required for Having Copies. 436 | 437 | You are not required to accept this License in order to receive or 438 | run a copy of the Program. Ancillary propagation of a covered work 439 | occurring solely as a consequence of using peer-to-peer transmission 440 | to receive a copy likewise does not require acceptance. However, 441 | nothing other than this License grants you permission to propagate or 442 | modify any covered work. These actions infringe copyright if you do 443 | not accept this License. Therefore, by modifying or propagating a 444 | covered work, you indicate your acceptance of this License to do so. 445 | 446 | 10. Automatic Licensing of Downstream Recipients. 447 | 448 | Each time you convey a covered work, the recipient automatically 449 | receives a license from the original licensors, to run, modify and 450 | propagate that work, subject to this License. You are not responsible 451 | for enforcing compliance by third parties with this License. 452 | 453 | An "entity transaction" is a transaction transferring control of an 454 | organization, or substantially all assets of one, or subdividing an 455 | organization, or merging organizations. If propagation of a covered 456 | work results from an entity transaction, each party to that 457 | transaction who receives a copy of the work also receives whatever 458 | licenses to the work the party's predecessor in interest had or could 459 | give under the previous paragraph, plus a right to possession of the 460 | Corresponding Source of the work from the predecessor in interest, if 461 | the predecessor has it or can get it with reasonable efforts. 462 | 463 | You may not impose any further restrictions on the exercise of the 464 | rights granted or affirmed under this License. For example, you may 465 | not impose a license fee, royalty, or other charge for exercise of 466 | rights granted under this License, and you may not initiate litigation 467 | (including a cross-claim or counterclaim in a lawsuit) alleging that 468 | any patent claim is infringed by making, using, selling, offering for 469 | sale, or importing the Program or any portion of it. 470 | 471 | 11. Patents. 472 | 473 | A "contributor" is a copyright holder who authorizes use under this 474 | License of the Program or a work on which the Program is based. The 475 | work thus licensed is called the contributor's "contributor version". 476 | 477 | A contributor's "essential patent claims" are all patent claims 478 | owned or controlled by the contributor, whether already acquired or 479 | hereafter acquired, that would be infringed by some manner, permitted 480 | by this License, of making, using, or selling its contributor version, 481 | but do not include claims that would be infringed only as a 482 | consequence of further modification of the contributor version. For 483 | purposes of this definition, "control" includes the right to grant 484 | patent sublicenses in a manner consistent with the requirements of 485 | this License. 486 | 487 | Each contributor grants you a non-exclusive, worldwide, royalty-free 488 | patent license under the contributor's essential patent claims, to 489 | make, use, sell, offer for sale, import and otherwise run, modify and 490 | propagate the contents of its contributor version. 491 | 492 | In the following three paragraphs, a "patent license" is any express 493 | agreement or commitment, however denominated, not to enforce a patent 494 | (such as an express permission to practice a patent or covenant not to 495 | sue for patent infringement). To "grant" such a patent license to a 496 | party means to make such an agreement or commitment not to enforce a 497 | patent against the party. 498 | 499 | If you convey a covered work, knowingly relying on a patent license, 500 | and the Corresponding Source of the work is not available for anyone 501 | to copy, free of charge and under the terms of this License, through a 502 | publicly available network server or other readily accessible means, 503 | then you must either (1) cause the Corresponding Source to be so 504 | available, or (2) arrange to deprive yourself of the benefit of the 505 | patent license for this particular work, or (3) arrange, in a manner 506 | consistent with the requirements of this License, to extend the patent 507 | license to downstream recipients. "Knowingly relying" means you have 508 | actual knowledge that, but for the patent license, your conveying the 509 | covered work in a country, or your recipient's use of the covered work 510 | in a country, would infringe one or more identifiable patents in that 511 | country that you have reason to believe are valid. 512 | 513 | If, pursuant to or in connection with a single transaction or 514 | arrangement, you convey, or propagate by procuring conveyance of, a 515 | covered work, and grant a patent license to some of the parties 516 | receiving the covered work authorizing them to use, propagate, modify 517 | or convey a specific copy of the covered work, then the patent license 518 | you grant is automatically extended to all recipients of the covered 519 | work and works based on it. 520 | 521 | A patent license is "discriminatory" if it does not include within 522 | the scope of its coverage, prohibits the exercise of, or is 523 | conditioned on the non-exercise of one or more of the rights that are 524 | specifically granted under this License. You may not convey a covered 525 | work if you are a party to an arrangement with a third party that is 526 | in the business of distributing software, under which you make payment 527 | to the third party based on the extent of your activity of conveying 528 | the work, and under which the third party grants, to any of the 529 | parties who would receive the covered work from you, a discriminatory 530 | patent license (a) in connection with copies of the covered work 531 | conveyed by you (or copies made from those copies), or (b) primarily 532 | for and in connection with specific products or compilations that 533 | contain the covered work, unless you entered into that arrangement, 534 | or that patent license was granted, prior to 28 March 2007. 535 | 536 | Nothing in this License shall be construed as excluding or limiting 537 | any implied license or other defenses to infringement that may 538 | otherwise be available to you under applicable patent law. 539 | 540 | 12. No Surrender of Others' Freedom. 541 | 542 | If conditions are imposed on you (whether by court order, agreement or 543 | otherwise) that contradict the conditions of this License, they do not 544 | excuse you from the conditions of this License. If you cannot convey a 545 | covered work so as to satisfy simultaneously your obligations under this 546 | License and any other pertinent obligations, then as a consequence you may 547 | not convey it at all. For example, if you agree to terms that obligate you 548 | to collect a royalty for further conveying from those to whom you convey 549 | the Program, the only way you could satisfy both those terms and this 550 | License would be to refrain entirely from conveying the Program. 551 | 552 | 13. Use with the GNU Affero General Public License. 553 | 554 | Notwithstanding any other provision of this License, you have 555 | permission to link or combine any covered work with a work licensed 556 | under version 3 of the GNU Affero General Public License into a single 557 | combined work, and to convey the resulting work. The terms of this 558 | License will continue to apply to the part which is the covered work, 559 | but the special requirements of the GNU Affero General Public License, 560 | section 13, concerning interaction through a network will apply to the 561 | combination as such. 562 | 563 | 14. Revised Versions of this License. 564 | 565 | The Free Software Foundation may publish revised and/or new versions of 566 | the GNU General Public License from time to time. Such new versions will 567 | be similar in spirit to the present version, but may differ in detail to 568 | address new problems or concerns. 569 | 570 | Each version is given a distinguishing version number. If the 571 | Program specifies that a certain numbered version of the GNU General 572 | Public License "or any later version" applies to it, you have the 573 | option of following the terms and conditions either of that numbered 574 | version or of any later version published by the Free Software 575 | Foundation. If the Program does not specify a version number of the 576 | GNU General Public License, you may choose any version ever published 577 | by the Free Software Foundation. 578 | 579 | If the Program specifies that a proxy can decide which future 580 | versions of the GNU General Public License can be used, that proxy's 581 | public statement of acceptance of a version permanently authorizes you 582 | to choose that version for the Program. 583 | 584 | Later license versions may give you additional or different 585 | permissions. However, no additional obligations are imposed on any 586 | author or copyright holder as a result of your choosing to follow a 587 | later version. 588 | 589 | 15. Disclaimer of Warranty. 590 | 591 | THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY 592 | APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT 593 | HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY 594 | OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, 595 | THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 596 | PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM 597 | IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF 598 | ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 599 | 600 | 16. Limitation of Liability. 601 | 602 | IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING 603 | WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS 604 | THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY 605 | GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE 606 | USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF 607 | DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD 608 | PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), 609 | EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF 610 | SUCH DAMAGES. 611 | 612 | 17. Interpretation of Sections 15 and 16. 613 | 614 | If the disclaimer of warranty and limitation of liability provided 615 | above cannot be given local legal effect according to their terms, 616 | reviewing courts shall apply local law that most closely approximates 617 | an absolute waiver of all civil liability in connection with the 618 | Program, unless a warranty or assumption of liability accompanies a 619 | copy of the Program in return for a fee. 620 | 621 | END OF TERMS AND CONDITIONS 622 | 623 | How to Apply These Terms to Your New Programs 624 | 625 | If you develop a new program, and you want it to be of the greatest 626 | possible use to the public, the best way to achieve this is to make it 627 | free software which everyone can redistribute and change under these terms. 628 | 629 | To do so, attach the following notices to the program. It is safest 630 | to attach them to the start of each source file to most effectively 631 | state the exclusion of warranty; and each file should have at least 632 | the "copyright" line and a pointer to where the full notice is found. 633 | 634 | 635 | Copyright (C) 636 | 637 | This program is free software: you can redistribute it and/or modify 638 | it under the terms of the GNU General Public License as published by 639 | the Free Software Foundation, either version 3 of the License, or 640 | (at your option) any later version. 641 | 642 | This program is distributed in the hope that it will be useful, 643 | but WITHOUT ANY WARRANTY; without even the implied warranty of 644 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 645 | GNU General Public License for more details. 646 | 647 | You should have received a copy of the GNU General Public License 648 | along with this program. If not, see . 649 | 650 | Also add information on how to contact you by electronic and paper mail. 651 | 652 | If the program does terminal interaction, make it output a short 653 | notice like this when it starts in an interactive mode: 654 | 655 | Copyright (C) 656 | This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'. 657 | This is free software, and you are welcome to redistribute it 658 | under certain conditions; type `show c' for details. 659 | 660 | The hypothetical commands `show w' and `show c' should show the appropriate 661 | parts of the General Public License. Of course, your program's commands 662 | might be different; for a GUI interface, you would use an "about box". 663 | 664 | You should also get your employer (if you work as a programmer) or school, 665 | if any, to sign a "copyright disclaimer" for the program, if necessary. 666 | For more information on this, and how to apply and follow the GNU GPL, see 667 | . 668 | 669 | The GNU General Public License does not permit incorporating your program 670 | into proprietary programs. If your program is a subroutine library, you 671 | may consider it more useful to permit linking proprietary applications with 672 | the library. If this is what you want to do, use the GNU Lesser General 673 | Public License instead of this License. But first, please read 674 | . 675 | --------------------------------------------------------------------------------