Unfolded Deep Kernel Estimation for Blind Image Super-resolution

Hongyi Zheng, Hongwei Yong, Lei Zhang, "Unfolded Deep Kernel Estimation for Blind Image Super-resolution". Accepted by ECCV2022.

[paper] [supp]

The implementation of UDKE is based on the awesome Image Restoration Toolbox [KAIR].

Requirement

• PyTorch 1.9+
• prettytable
• tqdm

Testing

Step 1

• Download testing kernels from [OneDrive].
• Unzip downloaded testing kernels and put the folders into ./kernels/test
• Download pretrained models from [OneDrive].
• Unzip downloaded file and put the folders into ./release/udke

Step 2

Configure options/test_udke.json. Important settings:

• task: task name.
• path/root: path to save the tasks.
• path/pretrained_netG: path to the folder containing the pretrained models.
• data/test/sigma: noise level
• data/test/sf: scale factor
• data/test/dataroot_h: path to testing sets

Step 3

python test_udke.py

Training

Step 1

Configure options/train_udke.json. Important settings:

• task: task name.
• path/root: path to save the tasks.
• data/train/sigma: noise level
• data/train/sf: scale factor
• data/train/dataroot_h: path to traing sets
• data/test/sigma: noise level
• data/test/sf: scale factor
• data/test/dataroot_h: path to testing sets

Step 2

python train_udke.py