Boosting Image and Video Compression via Learning Latent Residual Patterns (BMVC 2020)

Yen-Chung Chen, Keng-Jui Chang, Yi-Hsuan Tsai, Wei-Chen Chiu

British Machine Vision Conference (BMVC), 2020

[Paper][Video][Supplementary]

Introduction

The proposed framework aims to perform frame-by-frame video compression enhancement. We focus on utilizing the residual information, which is the difference between a compressed video and its corresponding original/uncompressed one, and propose a fairly efficient way to transmit the residual with the compressed video in order to boost the quality of video compression. For more details, please check out our video, paper, and supplementary materials.

Installation

Download repository:

git clone https://github.com/YenchungChen/Learning-Latent-Residual.git

Requirements are listed in environment.yml file.

Create the environment from the environment.yml using Anaconda, and activate it:

cd Learning-Latent-Residual
conda env create -f environment.yml
conda activate latent

Training

python train.py --config_filepath configs/train.yml

Testing

python eval.py --config_filepath configs/test.yml

Citation

If you find the code useful for your research, please cite:

@inproceedings{chen20bmvc,
 title = {Boosting Image and Video Compression via Learning Latent Residual Patterns},
 author = {Yen-Chung Chen and Keng-Jui Chang and Yi-Hsuan Tsai and Wei-Chen Chiu},
 booktitle = {British Machine Vision Conference (BMVC)},
 year = {2020}
}

@inproceedings{chen19clic,
 title = {Boosting Image and Video Compression via Learning Latent Residual Patterns},
 author = {Yen-Chung Chen and Keng-Jui Chang and Yi-Hsuan Tsai and Wei-Chen Chiu},
 booktitle = {Workshop and Challenge on Learned Image Compression (CLIC, in conjunction with CVPR)},
 year = {2019}
}