jingxu10 / MGANet-DCC2020

PyTorch implementation of "MGANet & DCC2020 paper"

MGANet

PyTorch implementation of "MGANet & DCC2020 paper". We have merged the codes of both methods.

• ArXiv
• Data Compression Conference (DCC2020)

The proposed MGANet framework

The guided encoder-decoder subnet

Installation

The code was developed using Python 3.6 & PyTorch 0.4 & CUDA 8.0. There may be a problem related to software versions. To fix the problem, you may look at the implementation in MGANet files and replace the syntax to match the new PyTorch environment.

Code v1.0

Currently, we release our research code for testing. It should produce the same results as in the paper under LD and AI configurations and frame number 3.

Test

• Pretrained models can be downloaded from this link!

https://drive.google.com/drive/folders/1xROBCUHgIX-3zqKMSHwKFEK8udIEicl1?usp=sharing

• It would be very easy to understand the test function and test on your own data.
• An example of test usage is shown as follows:

python MGANet_test_AI37.py --gpu_id 1 --is_training False

Train and Data

• We will update the training code and database for TUs' partition of compressed video by HEVC for training and better reading after recent paper deadline.

Video Results

• Here we provide quality enhancement results of compressed video for 18 widely used sequences for visual and quantitative comparisons.

Citation

If you use any part of our code, or our method is useful for your research, please consider citing:

@article{MGANET,
  author={Xiandong, Meng and Xuan, Deng and Shuyuan, Zhu and Shuaicheng, Liu and Chuan, Wang and Chen, Chen and Bing, Zeng},
  title={MGANet: A Robust Model for Quality Enhancement of Compressed Video},
  journal={arXiv:1811.09150}
}

@inproceedings{Meng_DCC2020,
  author={Xiandong, Meng and Xuan, Deng and Shuyuan, Zhu and Shuaicheng and Bing, Zeng},
  title={Flow-Guided Temporal-Spatial Network for HEVC Compressed Video Quality Enhancement},
  booktitle={Data Compression Conference(DCC)},
  year={2020}
}

Contact

We are glad to hear if you have any suggestions and questions. Please send email to [e-mail]