HUPE

This is an implementation of the paper, “HUPE: Heuristic Underwater Perceptual Enhancement with Semantic Collaborative Learning", Zengxi Zhang, Zhiying Jiang, Long Ma, Jinyuan Liu, Xin Fan, Risheng Liu*, 2025, International Journal of Computer Vision.

Overview

Download

Download the pre-trained model and put it in experiments/

• Baidu Yun
  code: ​ 1ecu ​
• Google Drive

Quick Run

To test the pre-trained models for Underwater Enhancement on example images:

Step 1: Generated priors from with underwater images:

run file 'IR_GDCP_forfolder.m' from root 'code/Generalization-of-the-Dark-Channel-Prior-for-Single-Image-Restoration-main/'

Step 2: Generated enhancement images:

cd code
python test.py

Contact

If you find our work useful in your research, please cite our conference version:

@inproceedings{zhang2023waterflow,
	title={WaterFlow: Heuristic Normalizing Flow for Underwater Image Enhancement and Beyond},
	author={Zhang, Zengxi and Jiang, Zhiying and Liu, Jinyuan and Fan, Xin and Liu, Risheng},
	booktitle={Proceedings of the 31st ACM International Conference on Multimedia},
	pages={7314--7323},
	year={2023}
}

Contact

Should you have any question, please contact [Zengxi Zhang].