sunny2109 / ShuffleMixer

[NeurIPS 2022] ShuffleMixer: An Efficient ConvNet for Image Super-Resolution

ShuffleMixer

Paper | Supplementary Material | Discussion

ShuffleMixer: An Efficient ConvNet for Image Super-Resolution

By Long Sun, Jinshan Pan, and Jinhui Tang

Network Architecture

Dependencies

• Linux (Tested on Ubuntu 18.04)
• Python 3.8.5 (Recommend to use Anaconda)
• PyTorch 1.11.0: pip install torch==1.11.0+cu113 torchvision==0.12.0+cu113 --extra-index-url https://download.pytorch.org/whl/cu113
• einops pip install einops
• fvcore pip install -U fvcore

Installation

# Clone the repo
git clone https://github.com/sunny2109/ShuffleMixer.git
# Install dependent packages
cd ShuffleMixer
pip install -r requirements.txt
# Install BasicSR
python setup.py develop

You can also refer to this INSTALL.md for installation

Training

• Run the following commands for training:

python basicsr/train.py -opt options/train/ShuffleMixer/train_base_DF2K_x4.yml

Testing

• Download the pretrained models.
• Download the testing dataset.
• Run the following commands:

python basicsr/test.py -opt options/test/ShuffleMixer/test_base_benchmark_x4.yml

• The test results will be in './results'.

Results

• Pretrained models and benchmark results can be downloaded from [Google Drive] or [Baidu Drive](code: idtn).

Citation

If you find this repository helpful, you may cite:

@InProceedings{Sun_2022,
    author    = {Sun, Long and Pan, Jinshan and Tang, Jinhui},
    title     = {{ShuffleMixer}: An Efficient ConvNet for Image Super-Resolution},
    booktitle = {Advances in Neural Information Processing Systems},
    year      = {2022}
}

Acknowledgment: This code is based on the BasicSR toolbox