STAMF

This project provides the code and datasets for 'STAMF: Synergistic Transformer and Mamba Fusion Network for RGB-Polarization based Underwater Salient Object Detection'

Dataset: TJUP-USOD and USODP: Baidu Pan：https://pan.baidu.com/s/1YjRaEdLTKTAggES3v1ziuA password：mmqw

You can retrain the model by running train_test_eval.py (with the Training parameter set to True), or load our provided weight files and run train_test_eval.py (with the Testing parameter set to True) to obtain the predicted images.

Thank you very much for your attention. We will continue to provide updates...

If you use our dataset or method, we would appreciate your citation:

@article{ma2025stamf,
  title={STAMF: Synergistic transformer and mamba fusion network for RGB-Polarization based underwater salient object detection},
  author={Ma, Qianwen and Li, Xiaobo and Li, Bincheng and Zhu, Zhen and Wu, Jing and Huang, Feng and Hu, Haofeng},
  journal={Information Fusion},
  pages={103182},
  year={2025},
  publisher={Elsevier}
}