CREStereo Dataset is a synthetic dataset that contains 200000 pairs of images and the corresponding dense disparity map.
Dataset can be downloaded at the following website: https://github.com/megvii-research/CREStereo/blob/master/README.md
The directory structure should be:
CREStereoData
├── hole
| ├── 0003c2ad-2128-4916-b450-60c22826bc83_left.jpg
| ├── 0003c2ad-2128-4916-b450-60c22826bc83_right.jpg
| ├── 0003c2ad-2128-4916-b450-60c22826bc83_left.disp.png
| ├── 0003c2ad-2128-4916-b450-60c22826bc83_right.disp.png
...
├── reflective
| |
...
├── shapenet
| |
...
├── tree
| |
...
Optionally you can write your own txt file and use all the parts of the dataset.
@inproceedings{li2022practical,
title={Practical stereo matching via cascaded recurrent network with adaptive correlation},
author={Li, Jiankun and Wang, Peisen and Xiong, Pengfei and Cai, Tao and Yan, Ziwei and Yang, Lei and Liu, Jiangyu and Fan, Haoqiang and Liu, Shuaicheng},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={16263--16272},
year={2022}
}