Home

Awesome

Revisiting Temporal Alignment for Video Restoration CVPR-2022 <br>

Kun Zhou, Wenbo Li, Liying Lu, Xiaoguang Han, Jiangbo Lu

[paper] <br>

We have provided the source code of our video super-resolution, video deblurring, and video denoising models. <br>

We provide our results at Google Cloud. <br>

The pre-trained models are uploaded in the google cloud. <br>

Some in-the-wild testing sequences are available here. <br>

File Structure <br>

<details> <summary>Click to expand </summary>

libs <br>

DcNv2 <br>

utils <br>

common.py <br> core.py <br> model_opr.py <br>

models <br>

VDB <br>

config.py <br> network.py <br> validate.py <br> sequence_test.py <br> load_VDB_Data.py <br> VideoDeblur.py <br>

VDN <br>

config.py <br> network.py <br> validate.py <br> validate_davis.py <br> sequence_test.py <br>

VSR_REDS <br>

config.py <br> network.py <br> validate.py <br>

VSR_VIMEO90K <br>

config.py <br> network.py <br> validate.py <br> sequence_test.py <br>

</details>

Usage

The DCNv2 should be installed correctly by running: <br> mask.sh in ./libs/DCNv2_latest/ <br> For evaluating the results of each model, you can run the corresponding "validate.py". <br> Also you can run the sequence_test.py for testing your own video sequences. <br>

Citing

If you find this code useful for your research, please consider citing the following paper:

@inproceedings{zhou2021rta, 
  title={Revisiting Temporal Alignment for Video Restoration},
  author={Kun Zhou and Wenbo Li and Liying Lu and Xiaoguang Han and Jiangbo Lu}, 
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition} 
  year={2022} 
}

License

Our code is for research purposes only.