DE-YOLO

PyTorch code for our ACCV2022 paper "DENet: Detection-driven Enhancement Network for Object Detection under Adverse Weather Conditions"

Dependencies

• python==3.7.5
• torch==1.7.1
• torchvision==0.8.2
• tensorboard==2.5.0
• numpy==1.19.5
• opencv-python==4.2.0.34

cd DE-YOLO 
pip install -r ./requirements.txt

Datasets and Models

Please download the processed datasets and pretrained models from the anonymous Github links below.

RTTS ExDark Pretrained Models

Folder structure

Download the datasets and pretrained models first. Please prepare the basic folder structure as follows.

/parent_folder
  /datasets   # folder for datasets 
    /RTTS
    /ExDark
    ...
  /DE-YOLO
    /data     # config files for datasets
    /models   # python files for DE-YOLO
    /pretrained_models  # folder for pretrained models
    requirements.txt
    README.md
    ...

Quick Test

Evaluation on real-world low-light images from ExDark

# put datasets and pretrained model in the corresponding directory 
cd DE-YOLO 
bash test_exdark_deyolo.sh

Evaluation on natural foggy images from RTTS

# put datasets and pretrained model in the corresponding directory
cd DE-YOLO 
bash test_rtts_deyolo.sh

Train

The source code for training our DE-YOLO will be available after the publication of the paper.