Stereo Video Super-Resolution via Exploiting View-Temporal Correlations, In ACM MM 2021.
Ruikang Xu, Zeyu Xiao, Mingde Yao, Yueyi Zhang, Zhiwei Xiong.

Paper|Supplemental Material|Video

• This repository is based on [EDVR/old_version], you can install DeformConv by following [EDVR/old_version]
• Python 3 (Recommend to use Anaconda)
• PyTorch 1.2.0: conda install pytorch=1.2.0 torchvision cudatoolkit=9.2 -c pytorch
• numpy: pip install numpy
• opencv: pip install opencv-python
• tensorboardX: pip install tensorboardX

• The SceneFlow dataset can be downloaded from this link.
• The KITTI-2012 dataset can be downloaded from this link.
• The KITTI-2015 dataset can be downloaded from this link.

We take the sceneflow dataset as an example:

• Prepare text files for loading data:

cd dataPrepare && python creatTxt_sceneflow.py

• Train the model:

cd code && python train.py

• Test the model:

cd code && python test.py

@inproceedings{xu2021stereo,
  title={Stereo video super-resolution via exploiting view-temporal correlations},
  author={Xu, Ruikang and Xiao, Zeyu and Yao, Mingde and Zhang, Yueyi and Xiong, Zhiwei},
  booktitle={Proceedings of the 29th ACM International Conference on Multimedia},
  pages={460--468},
  year={2021}
}

Any question regarding this work can be addressed to [email protected].