This is a pytorch implementation of Bridging Component Learning with Degradation Modelling for Blind Image Super-Resolution. This repo is built on the basis of DAN and BasicSR, thanks for their open-sourcing!
- python3
- NVIDIA GPU + CUDA
- pytorch >= 1.7.1
- python packages:
pip install -r requirements.txt - bascisr:
python setup.py develop
Download the DIV2K and Flickr2K and merge it into one folder. Modify options/train/train_setting.yml and run the following command
python basicsr/train.py -opt=options/train/train_setting.yml
There are two blind settings mentioned in our paper. For setting1, we synthesize the Gaussian8 datasets with five datasets: Set5, Set14, BSD100, Urban100, Manga109. Please refer to this repository for more details.
For setting2, we using the benchmark dataset DIV2KRK from KernelGAN.
The pretrained models can be downloaded here (setting2 x2 model is missed and we are re-training it).
Modify the dataset path and test settings in options/test/test_setting.yml and run the following command
python basicsr/test.py -opt=options/test/test_setting.yml
If you find this repo useful, please consider citing our work:
@ARTICLE{9925720,
author={Wu, Yixuan and Li, Feng and Bai, Huihui and Lin, Weisi and Cong, Runmin and Zhao, Yao},
journal={IEEE Transactions on Multimedia},
title={Bridging Component Learning with Degradation Modelling for Blind Image Super-Resolution},
year={2022},
volume={},
number={},
pages={1-16},
doi={10.1109/TMM.2022.3216115}}