/IKM

codes for IKM (arXiv2021, Submitted to IEEE Trans)

Primary LanguagePython

Image-specific Convolutional Kernel Modulation for Single Image Super-resolution

This repository is for IKM introduced in the following paper

Yuanfei Huang, Jie Li, Yanting Hu, Hua Huang and Xinbo Gao*, "Image-specific Convolutional Kernel Modulation for Single Image Super-resolution", arXiv preprint arXiv:2111.08362(2021)

arXiv

Overflow

Pipeline of IKM

Framework of UHDN

Dependenices

  • python 3.8
  • pytorch >= 1.7.0
  • NVIDIA GPU + CUDA

Data preparing

Download DIV2K and Flickr2K datasets into the path "../../Datasets/Train/DF2K".

Train

  1. Replace the train dataset path '../../Datasets/Train/' and validation dataset '../../Datasets/Test/' with your training and validation datasets, respectively.

  2. Set the configurations in 'option.py' as you want.

python main.py --train 'Train'

Test

  1. Download models from Google Drive or BaiduYun(password: 06v3).

  2. Replace the test dataset path '../../Datasets/Test/' with your datasets.

python main.py --train 'Test'

Results

Quantitative Results (PSNR/SSIM)

Quantitative Results

Qualitative Results

Fig.6

Citation

@misc{huang2021imagespecific,
      title={Image-specific Convolutional Kernel Modulation for Single Image Super-resolution}, 
      author={Yuanfei Huang and Jie Li and Yanting Hu and Xinbo Gao and Hua Huang},
      year={2021},
      eprint={2111.08362},
      archivePrefix={arXiv},
      primaryClass={eess.IV}
}