Official implementation for TGRS paper : “Receptive-field and Direction Induced Attention Network for Infrared Dim Small Target Detection with a Large-scale Dataset IRDST”
Please cite our paper in your publications if our work helps your research. BibTeX reference is as follows.
@article{TGRS23RDIAN,
author={Sun, Heng and Bai, Junxiang and Yang, Fan and Bai, Xiangzhi},
journal={IEEE Transactions on Geoscience and Remote Sensing},
title={Receptive-Field and Direction Induced Attention Network for Infrared Dim Small Target Detection With a Large-Scale Dataset IRDST},
year={2023},
volume={61},
number={},
pages={1-13},
doi={10.1109/TGRS.2023.3235150}
}
And if the implementation of this repo is helpful to you, just star it.
- Python 3.9
- Pytorch 1.8.0
The IRDST dataset: http://xzbai.buaa.edu.cn/datasets.html
Currently, the single-frame version of the IRDST dataset has been open-sourced. The sequence version of the IRDST dataset is still being checked and will be updated in the future.
The training and testing experiments are conducted using PyTorch with a single GeForce RTX 3090 GPU of 24 GB Memory.
The trained model params are in ./params
python train.py
python test.py
MIT License