Skip to content

STDR: Spatio-Temporal Detail Information Retrieval for Compressed Video Quality Enhancement

License

Notifications You must be signed in to change notification settings

xiaomingxige/STDR

Repository files navigation

Spatio-Temporal Detail Information Retrieval for Compressed Video Quality Enhancement (TMM 2023)

The PyTorch implementation for the STDR: Spatio-Temporal Detail Information Retrieval for Compressed Video Quality Enhancement which is accepted by [IEEE TMM 2023].

1. Pre-request

1.1. Environment

conda create -n STDR python=3.10.11
conda activate STDR
conda install pytorch==1.12.0 torchvision==0.13.0 torchaudio==0.12.0 cudatoolkit=11.3 -c pytorch

git clone --depth=1 https://github.com/xiaomingxige/STDR
cd STDR
pip install -r requirements.txt

1.2. DCNv2

Build DCNv2

cd ops/dcn/
# You may need to modify the paths of cuda before compiling.
bash build.sh

Check if DCNv2 works (optional)

python simple_check.py

The DCNv2 source files here is different from the open-sourced version due to incompatibility. [issue]

1.3. MFQEv2 dataset

1.3.1 Download and compress videos

Please check here.

1.3.2 Edit YML

You should modify option_mfqev2_1G.yml according to your environment

1G: 1 GPUs will be used for the below training.

You can also edit option_mfqev2_2G.yml and option_mfqev2_4G.yml if needed.

1.3.3 Generate LMDB

We now generate LMDB to speed up IO during training.

python create_lmdb_mfqev2.py --opt_path option_mfqev2_1G.yml

Finally, the MFQEv2 dataset root will be sym-linked to the folder ./data/ automatically.

So that we and programmes can access MFQEv2 dataset at ./data/ directly.

2. Train

See script.sh.

3. Test

We utilize 1 NVIDIA GeForce RTX 3090 GPU for testing:

python test_video.py

Citation

If you find this project is useful for your research, please cite:

@article{luo2023spatio,
  title={Spatio-Temporal Detail Information Retrieval for Compressed Video Quality Enhancement},
  author={Luo, Dengyan and Ye, Mao and Li, Shuai and Zhu, Ce and Li, Xue},
  journal={IEEE Transactions on Multimedia},
  volume={25},
  pages={6808--6820},
  year={2023},
  publisher={IEEE}
}

Acknowledgements

This work is based on STDF-Pytoch. Thank RyanXingQL for sharing the codes.

About

STDR: Spatio-Temporal Detail Information Retrieval for Compressed Video Quality Enhancement

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published