Rui Lin(Github:JacklE0niden)
baseline:DM2F By Zijun Deng, Lei Zhu, Xiaowei Hu, Chi-Wing Fu, Xuemiao Xu, Qing Zhang, Jing Qin, and Pheng-Ann Heng.
This repo is the implementation of "Deep Multi-Model Fusion for Single-Image Dehazing" (ICCV 2019), written by Zijun Deng at the South China University of Technology.
Some basic dehazing results can be found at Report.md
- Download model weight files:https://pan.baidu.com/s/1gnxMVDx2S6QHSJdwG9FX0Q?pwd=8nfj
Make sure you have Python>=3.7
installed on your machine.
Environment setup:
-
Create conda environment
conda create -n dm2f conda activate dm2f
-
Install dependencies (test with PyTorch 1.8.0):
-
Install pytorch==1.8.0 torchvision==0.9.0 (via conda, recommend).
-
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.
-
Make a directory
./data
and create a symbolic link for uncompressed data, e.g.,./data/RESIDE
.
-
- Run by
python tools/preprocess_ohaze_data.py
Set the path of pretrained ResNeXt model in resnext/config.py- Set the path of datasets in tools/config.py
- Run by
train.sh
- Set the path of five benchmark datasets in tools/config.py.
- Put the trained model in
./ckpt/
. - Run by
test.sh
Settings of testing were set at the top of test.py
, and you can conveniently
change them as you need.
-DM2F -dataset.py -train_1_baseline.py -data -O-HAZE -# O-HAZY NTIRE 2018 -GT -HAZY