Pytorch implementation of ZID (TIP 2020) [paper]

• Python == 3.6.10
• Pytorch == 1.1.0
• opencv-python == 3.4.2.16
• opencv-contrib-python == 3.4.2.16

We also export our conda virtual environment as ZID.yaml. You can use the following command to create the environment.

conda env create -f ZID.yaml

You can use the following command to dehaze the test image in ./data:

python dehazing.py

If you want to test ZID on a real world image which does not have ground truth. You can use the following command:

python RW_dehazing.py

The only difference between two command is whether the program calculates PSNR and SSIM.

If you find ZID useful in your research, please consider citing:

@article{ZID,
author = {Li, Boyun and Gou, Yuanbiao and Liu, Jerry Zitao and Zhu, Hongyuan and Zhou, Joey Tianyi and Peng, Xi},
title = {{Zero-Shot Image Dehazing}},
journal = {IEEE Transactions on Image Processing},
year = {2020},
volume = {29},
pages = {8457--8466},
month = aug
}