IRSTD-Benchmark

Paper

• IR-TransDet: Infrared Dim and Small Target Detection With IR-Transformer
• Infrared dim and small target detection based on U-Transformer

简体中文   Project

This repository is a benchmark tool for weak small target detection, aiming to provide a platform that integrates typical weak small target detection algorithms and unified evaluation metrics. This tool can help researchers and developers evaluate the performance of different algorithms on weak small target detection tasks.

Features

• Provides standard evaluation metrics for weak small target detection to objectively evaluate algorithm performance, including but not limited to:
  • Precision-Recall
  • F1 score
• Simple and easy-to-use interface and sample code for users to quickly get started and integrate their own algorithms or conduct evaluation experiments.

IRSTD1K	MDFA

Merged	SirstAUG

Dataset

Dataset	Link
Sirst AUG	Link
MDFA	Link
IRSTD1K	Link

The above datasets have been organized and uploaded to Google Drive, and users can download and use them directly.

Installation and Usage

Install Matlab Python API

pip install -r requirements.txt

Usage

python method.py --dataset mdfa > mdfa.txt
                            irstd > istd1k.txt
                            sirstaug > sirstaug.txt
                            merged > merged.txt 

Plot ROC curve

python plot_curve.py

Add custom model

Please refer to model.md for add custom model.

Contribution

Contributions to this repository are welcome! If you have new algorithm implementations, improvements, or suggestions for other feature enhancements, please submit a pull request. We are happy to accept new contributions and continuously improve this tool.

Cite

@article{lin2022infrared,
  title={Infrared dim and small target detection based on U-Transformer},
  author={Lin, Jian and Zhang, Kai and Yang, Xi and Cheng, Xiangzheng and Li, Chenhui},
  journal={Journal of Visual Communication and Image Representation},
  pages={103684},
  year={2022},
  publisher={Elsevier}
}

@ARTICLE{10295542,
  author={Lin, Jian and Li, Shaoyi and Zhang, Liang and Yang, Xi and Yan, Binbin and Meng, Zhongjie},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={IR-TransDet: Infrared Dim and Small Target Detection With IR-Transformer}, 
  year={2023},
  volume={61},
  number={},
  pages={1-13},
  doi={10.1109/TGRS.2023.3327317}}