Uncertainty Inspired Underwater Image Enhancement (ECCV 2022)(Paper)

The Pytorch Implementation of ''Uncertainty Inspired Underwater Image Enhancement''.

Introduction

In this project, we use Ubuntu 16.04.5, Python 3.7, Pytorch 1.7.1 and one NVIDIA RTX 2080Ti GPU.

Running

Testing

Download the pretrained model pretrained model.

Check the model and image pathes in Test_MC.py and Test_MP.py, and then run:

python Test_MC.py

python Test_MP.py

Training

To train the model, you need to prepare our dataset.

Check the dataset path in Train.py, and then run:

python Train.py

Citation

If you find PUIE-Net is useful in your research, please cite our paper:

@inproceedings{Fu_2022, 
	title={Uncertainty Inspired Underwater Image Enhancement}, 
	author={Fu, Zhenqi and Wang, Wu and Huang, Yue and Ding, Xinghao and Ma, Kai-Kuang}, 
	booktitle={European Conference on Computer Vision (ECCV)},
	year={2022}, 
	pages={465--482},	
}