Omni-Kernel Network for Image Restoration [AAAI-24]

Installation

The project is built with PyTorch 3.8, PyTorch 1.8.1. CUDA 10.2, cuDNN 7.6.5 For installing, follow these instructions:

conda install pytorch=1.8.1 torchvision=0.9.1 -c pytorch
pip install tensorboard einops scikit-image pytorch_msssim opencv-python

Install warmup scheduler:

cd pytorch-gradual-warmup-lr/
python setup.py install
cd ..

ITS FLOPs: 39.67G, Params: 4.72M Training and testing details can be found in the individual directories.

Images

Citation

@inproceedings{cui2024omni,
  title={Omni-Kernel Network for Image Restoration},
  author={Cui, Yuning and Ren, Wenqi and Knoll, Alois},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={38},
  number={2},
  pages={1426--1434},
  year={2024}
}

Contact

Should you have any question, please contact Yuning Cui.