By Yesian Rohn

Our Model——CL2S means "Change Logarithmic to Sinusoidal function".

Results on Benchmark：

Results on C-Haze：

The model weights can be found at Google Drive.

Make sure you have Python>=3.7 installed on your machine.

Environment setup:

1. Create conda environment

   conda create -n cl2s
   conda activate cl2s
   

2. Install dependencies (test with PyTorch 1.8.0):

   1. Install pytorch==1.8.0 torchvision==0.9.0 (via conda, recommend).

   2. Install other dependencies

      pip install -r requirements.txt
      

• Prepare the dataset

  • Download the RESIDE dataset from the official webpage.

  • Download the O-Haze dataset from the official webpage.

  • Download the HazeRD dataset from the Baidu Netdisk.

  • C-Haze is collected by me from the Internet.

  • Make a directory ./data and create a symbolic link for uncompressed data, e.g., ./data/RESIDE.

1. Set the path of datasets in tools/config.py
2. Run by python train.py

Use pretrained ResNeXt (resnext101_32x8d) from torchvision.

Hyper-parameters of training were set at the top of train.py, and you can conveniently change them as you need.

Training a model on a single RTX3090 GPU takes about 5 hours.

1. Set the path of five benchmark datasets in tools/config.py.
2. Put the trained model in ./ckpt/.
3. Run by python test.py

Settings of testing were set at the top of test.py, and you can conveniently change them as you need.

CL2S is released under the MIT license.

We gratefully acknowledge the work of Zijun Deng et al. presented in their ICCV 2019 paper, "Deep Multi-Model Fusion for Single-Image Dehazing". Our project is built upon the foundation laid by their research and the open-source code shared at DM2F-Net.

Citation Guidance:

For utilizing ideas or code from this project, please cite:

@inproceedings{deng2019deep,
  title={Deep multi-model fusion for single-image dehazing},
  authors={Deng, Zijun; Zhu, Lei; Hu, Xiaowei; Fu, Chi-Wing; Xu, Xuemiao; Zhang, Qing; Qin, Jing; Heng, Pheng-Ann},
  booktitle={IEEE/CVF International Conference on Computer Vision},
  pages={2453--2462},
  year={2019}
}

We appreciate the authors' contributions and encourage proper citations to recognize their original work.