[ICIP 2023] LLIEFormer: A Low-Light Image Enhancement Transformer Network with a Degraded Restoration Model

Paper | Code

LLIEFormer: A Low-Light Image Enhancement Transformer Network with a Degraded Restoration Model Xunpeng Yi, Yuxuan Wang, Yizhen Zhao, Jia Yan, Weixia Zhang in ICIP 2023

Pretrained Model

We provide the pretrained models:

• LLIEFormer pretrained on LOL dataset at [Google Drive] | [Baidu Drive] (code: 9m5g).

Method	PSNR	SSIM
LLIEFormer	22.08	0.883

• LLIEFormer pretrained on PairL1.6K dataset at [Google Drive] | [Baidu Drive] (code: v513) (update version).

Method	PSNR	SSIM
Zero-DCE	16.90	0.678
KinD	17.27	0.645
KinD++	18.52	0.701
LIME	18.19	0.671
RUAS	17.91	0.633
RetinexNet	16.71	0.626
LLIEFormer	25.14	0.797

Dataset

You can put and rename the dataset in the following way:

    dataset/
        LOLdataset/
            train/
                high/
                low/
            eval/
                high/
                low/

Test

cd LLIEFormer-main/
CUDA_VISIBLE_DEVICES=0 python test.py

Train

cd LLIEFormer-main/
CUDA_VISIBLE_DEVICES=0 python train.py

Citation

If you find our work useful for your research, please cite our paper

@inproceedings{yi2023llieformer,
  title={Llieformer: A Low-Light Image Enhancement Transformer Network with a Degraded Restoration Model},
  author={Yi, Xunpeng and Wang, Yuxuan and Zhao, Yizhen and Yan, Jia and Zhang, Weixia},
  booktitle={2023 IEEE International Conference on Image Processing (ICIP)},
  pages={1195--1199},
  year={2023},
  organization={IEEE}
}