EDPN: Enhanced Deep Pyramid Network for Blurry Image Restoration

Ruikang Xu, Zeyu Xiao, Jie Huang, Yueyi Zhang, Zhiwei Xiong. EDPN: Enhanced Deep Pyramid Network for Blurry Image Restoration. In CVPRW 2021. (Winner of NTIRE Challenge 2021 on Image Deblurring Track 1. Low Resolution)

arXiv | CVF

Dependencies

• This repository is based on [EDVR/old_version], you can install DeformConv by following [EDVR/old_version]
• Python 3 (Recommend to use Anaconda)
• PyTorch 1.2.0: conda install pytorch=1.2.0 torchvision cudatoolkit=9.2 -c pytorch
• numpy: pip install numpy
• opencv: pip install opencv-python
• tensorboardX: pip install tensorboardX

Prepare your data

Take blurry image super-resolution as an example

cd deblur_SR/datasets && python createTxt.py

Train the model

Take blurry image super-resolution as an example

cd deblur_SR && python train.py

Test the model

Take blurry image super-resolution as an example

cd deblur_SR && python inference.py

Dataset

CodaLab | NTIRE2021

Citation

@InProceedings{Xu_2021_CVPRW_EDPN,
    author = {Xu, Ruikang and Xiao, Zeyu and Huang, Jie and Zhang, Yueyi and Xiong, Zhiwei},
    title = {EDPN: Enhanced Deep Pyramid Network for Blurry Image Restoration},
    booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition WorkShops (CVPRW)},
    month = {June},
    year = {2021}
}  

@InProceedings{Nah_2021_CVPRW_Deblur,
    author = {Nah, Seungjun and Son, Sanghyun and Lee, Suyoung and Timofte, Radu and Lee, Kyoung Mu},
    title = {NTIRE 2021 Challenge on Image Deblurring},
    booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition WorkShops (CVPRW)},
    month = {June},
    year = {2021}
}

Contact

Any question regarding this work can be addressed to xurk@mail.ustc.edu.cn and zeyuxiao@mail.ustc.edu.cn.