🔥 OASNet: OASNet: Orthogonal Attention-Guided Spatial–Semantic Representation Learning Network for Infrared Small Target Detection [📄 Paper Link]

Xiaoyang Yuan, Chunling Yang, Yuze Li, Yan Zhang, IEEE Geoscience and Remote Sensing Letters 2025.

If the implementation of this repo is helpful to you, just star it！⭐⭐⭐

Usage

1. Data

• The SIRST dataset download dir [ACM]
• The NUDT-SIRST dataset download dir [DNANet]
• The IRSTD-1k dataset download dir [ISNet]

Our project has the following structure:

├──./dataset/
│    ├── IRSTD-1K
│    │    ├── images
│    │    │    ├── XDU0.png
│    │    │    ├── XDU1.png
│    │    │    ├── ...
│    │    ├── masks
│    │    │    ├── XDU0.png
│    │    │    ├── XDU1.png
│    │    │    ├── ...
│    │    ├── 80_20
│    │    │    ├── train.txt
│    │    │    ├── test.txt
│    ├── NUDT-SIRST
│    │    ├── images
│    │    │    ├── 000001.png
│    │    │    ├── 000002.png
│    │    │    ├── ...
│    │    ├── masks
│    │    │    ├── 000001.png
│    │    │    ├── 000002.png
│    │    │    ├── ...
│    │    ├── 80_20
│    │    │    ├── train.txt
│    │    │    ├── test.txt
│    ├── ...
│    ├── ...
│    ├── NUAA-SIRST
│    │    ├── images
│    │    │    ├── Misc_1.png
│    │    │    ├── Misc_2.png
│    │    │    ├── ...
│    │    ├── masks
│    │    │    ├── Misc_1.png
│    │    │    ├── Misc_2.png
│    │    │    ├── ...
│    │    ├── 80_20
│    │    │    ├── train.txt
│    │    │    ├── test.txt

2. Train.

python train_oas.py 

3. Test.

python test_oas.py 

• This code is highly borrowed from AMFU. Thanks to Won Young Chung.
• This code is highly borrowed from DNANet. Thanks to Boyang Li.

Citation

If you find the code useful, please consider citing our paper using the following BibTeX entry.

@ARTICLE{11219229,
  author={Yuan, Xiaoyang and Yang, Chunling and Li, Yuze and Zhang, Yan},
  journal={IEEE Geoscience and Remote Sensing Letters}, 
  title={OASNet: Orthogonal Attention-Guided Spatial–Semantic Representation Learning Network for Infrared Small Target Detection}, 
  year={2025},
  volume={22},
  number={},
  pages={1-5},
  doi={10.1109/LGRS.2025.3626394}}

Contact

Welcome to raise issues or email to yuanxiaoyang1998@outlook.com for any question regarding our OASNet.