WaveNet

This project provides the code and results for 'WaveNet: Wavelet Network With Knowledge Distillation for RGB-T Salient Object Detection', IEEE TIP, 2023. IEEE link

• Old codebase: https://github.com/nowander/WaveNet/tree/default

Requirements

Python 3.7+, Pytorch 1.5.0+, Cuda 10.2+, TensorboardX 2.1, opencv-python, pytorch_wavelets, timm.

Architecture and Details

Results

Preparation

• Download the RGB-T raw data from LSNet.
  
• Options: Download the pre-trained wavemlp-s from wavemlp.
  
• We have two ways of training knowledge distillation:
  1. Load the SwinNet model, please refer to the specific configuration of SwinNet.
  2. Directly load the prediction maps of SwinNet baidu pin: py5y.
     We use prediction maps of SwinNet as the default setting.

Training & Testing

Modify the train_root train_root save_path path in config.py according to your own data path.

• Train the WaveNet:

  python train.py

Modify the test_path path in config.py according to your own data path.

• Test the WaveNet:

  python test.py

Evaluate tools

• You can select one of the toolboxes to get the metrics CODToolbox / PySODMetrics

Saliency Maps

• RGB-T baidu pin: gl01
  

Pretraining Models

• RGB-T baidu pin: v5pb
  

Citation

    @ARTICLE{10127616,
        author={Zhou, Wujie and Sun, Fan and Jiang, Qiuping and Cong, Runmin and Hwang, Jenq-Neng},
        journal={IEEE Transactions on Image Processing}, 
        title={WaveNet: Wavelet Network With Knowledge Distillation for RGB-T Salient Object Detection}, 
        year={2023},
        volume={32},
        number={},
        pages={3027-3039},
        doi={10.1109/TIP.2023.3275538}}     

Acknowledgement

The implementation of this project is based on the codebases below.

• BBS-Net
  
• LSNet
  
• Wavemlp
  
• Evaluate tools CODToolbox / PySODMetrics
  

If you find this project helpful, Please also cite the codebases above. Besides, we also thank zyrant.

Contact

Please drop me an email for any problems or discussion: https://wujiezhou.github.io/ ([email protected]).