├── requirements.txt
└── README.md


/requirements.txt:
--------------------------------------------------------------------------------
 1 | argparse
 2 | easydict
 3 | matplotlib
 4 | numpy
 5 | opencv-python
 6 | pyexr
 7 | scipy
 8 | torch>=1.4.0
 9 | torchvision
10 | tensorboardX
11 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
 1 | # Stereo 3D Reconstruction
 2 | 
 3 | This repository contains the source code for the paper [Toward 3D Object Reconstruction from Stereo Images](https://arxiv.org/abs/1910.08223.pdf).
 4 | 
 5 | **Important Note:** The source code is in the ([Stereo2Voxel](https://github.com/hzxie/Stereo-3D-Reconstruction/tree/Stereo2Voxel)/[Stereo2Point](https://github.com/hzxie/Stereo-3D-Reconstruction/tree/Stereo2Point)) branches of the repository.
 6 | 
 7 | ![Overview](https://www.infinitescript.com/projects/Stereo-3D-Reconstruction/Stereo-3D-Reconstruction-Overview.jpg)
 8 | 
 9 | ## Cite this work
10 | 
11 | ```
12 | @article{xie2021towards,
13 |   title={Toward 3D Object Reconstruction from Stereo Images},
14 |   author={Xie, Haozhe and
15 |           Tong, Xiaojun and
16 |           Yao, Hongxun and
17 |           Zhou, Shangchen and
18 |           Zhang, Shengping and
19 |           Sun, Wenxiu},
20 |   journal={Neurocomputing},
21 |   year={2021}
22 | }
23 | ```
24 | 
25 | ## Datasets
26 | 
27 | We use the [StereoShapeNet](https://www.shapenet.org/) dataset in our experiments, which is available below:
28 | 
29 | - [StereoShapeNet](https://gateway.infinitescript.com/?fileName=StereoShapeNet)
30 | 
31 | ## Pretrained Models
32 | 
33 | The pretrained models on StereoShapeNet are available as follows:
34 | 
35 | - [Stereo2Voxel for StereoShapeNet](https://gateway.infinitescript.com/?fileName=Stereo2Voxel-StereoShapeNet.pth) (309 MB)
36 | - [Stereo2Point for StereoShapeNet](https://gateway.infinitescript.com/?fileName=Stereo2Point-StereoShapeNet.pth) (356 MB)
37 | 
38 | ## Prerequisites
39 | 
40 | #### Clone the Code Repository
41 | 
42 | ```
43 | git clone https://github.com/hzxie/Stereo-3D-Reconstruction.git
44 | ```
45 | 
46 | #### Install Python Denpendencies
47 | 
48 | ```
49 | cd Stereo-3D-Reconstruction
50 | pip install -r requirements.txt
51 | ```
52 | 
53 | #### Train/Test Stereo2Voxel
54 | 
55 | ```
56 | git checkout Stereo2Voxel
57 | ```
58 | 
59 | #### Train/Test Stereo2Point
60 | 
61 | ```
62 | git checkout Stereo2Point
63 | 
64 | cd extensions/chamfer_dist
65 | python setup.py install --user
66 | ```
67 | 
68 | #### Update Settings in `config.py`
69 | 
70 | You need to update the file path of the datasets:
71 | 
72 | ```
73 | __C.DATASETS.SHAPENET.LEFT_RENDERING_PATH   = '/path/to/ShapeNetStereoRendering/%s/%s/render_%02d_l.png'
74 | __C.DATASETS.SHAPENET.RIGHT_RENDERING_PATH  = '/path/to/ShapeNetStereoRendering/%s/%s/render_%02d_r.png'
75 | __C.DATASETS.SHAPENET.LEFT_DISP_PATH        = '/path/to/ShapeNetStereoRendering/%s/%s/disp_%02d_l.exr'
76 | __C.DATASETS.SHAPENET.RIGHT_DISP_PATH       = '/path/to/ShapeNetStereoRendering/%s/%s/disp_%02d_r.exr'
77 | __C.DATASETS.SHAPENET.VOLUME_PATH           = '/path/to/ShapeNetVox32/%s/%s.mat'
78 | ```
79 | 
80 | ## Get Started
81 | 
82 | To train GRNet, you can simply use the following command:
83 | 
84 | ```
85 | python3 runner.py
86 | ```
87 | 
88 | To test GRNet, you can use the following command:
89 | 
90 | ```
91 | python3 runner.py --test --weights=/path/to/pretrained/model.pth
92 | ```
93 | 
94 | ## License
95 | 
96 | This project is open sourced under MIT license.
97 | 


--------------------------------------------------------------------------------