DeepInfrared aims to be an open benchmark for infrared small target detection, currently consisting of:

1. Public infrared small target dataset (SIRST-V2);
2. Evaluation metrics specially designed (mNoCoAP);
3. An open source toolbox based on PyTorch (DeepInfrared).

Single-frame InfraRed Small Target (SIRST) detection has been a challenging task due to a lack of inherent characteristics, imprecise bounding box regression, a scarcity of real-world datasets, and sensitive localization evaluation. In this paper, we propose a comprehensive solution to these challenges. First, we find that the existing anchor-free label assignment method is prone to mislabeling small targets as background, leading to their omission by detectors. To overcome this issue, we propose an all-scale pseudo-box-based label assignment scheme that relaxes the constraints on scale and decouples the spatial assignment from the size of the ground-truth target. Second, motivated by the structured prior of feature pyramids, we introduce the one-stage cascade refinement network (OSCAR), which uses the high-level head as soft proposals for the low-level refinement head. This allows OSCAR to process the same target in a cascade coarse-to-fine manner. Finally, we present a new research benchmark for infrared small target detection, consisting of the SIRST-V2 dataset of real-world, high-resolution single-frame targets, the normalized contrast evaluation metric, and the DeepInfrared toolkit for detection. We conduct extensive ablation studies to evaluate the components of OSCAR and compare its performance to state-of-the-art model-driven and data-driven methods on the SIRST-V2 benchmark. Our results demonstrate that a top-down cascade refinement framework can improve the accuracy of infrared small target detection without sacrificing efficiency.

For details see OSCAR. The speed and accuracy are listed as follows:

As a part of the DeepInfrared Eco-system, we provide the SIRST-V2 dataset as a benchmark. SIRST-V2 is a dataset specially constructed for single-frame infrared small target detection, in which the images are selected from thousands of infrared sequences for different scenarios.

Annotation formats available:

• bounding box;
• semantic segmentation;
• normalized contrast (produced when data loading).

The dataset can be downloaded here.

Please refer to Installation for installation instructions.

# assume that you are under the root directory of this project,
# and you have activated your virtual environment if needed.
# and with SIRST-V2 dataset in 'data/sirst/'

python tools/train_det.py \
    configs/oscar/sota/oscar_w_noco_head_r18_caffe_fpn_p2_gn-head_1x_sirst_det2noco.py \
    --gpu-id 0 \
    --work-dir work_dirs/oscar_w_noco_head_r18_caffe_fpn_p2_gn-head_1x_sirst_det2noco

python tools/test_det.py \
    configs/oscar/sota/oscar_w_noco_head_r18_caffe_fpn_p2_gn-head_1x_sirst_det2noco.py \
    work_dirs/oscar_w_noco_head_r18_caffe_fpn_p2_gn-head_1x_sirst_det2noco/best.pth --eval "mNoCoAP"

For your convenience, we provide the following trained models.

For traditional methods, e.g., low-rank based or local contrast based approaches, we provide the predicted target images:

Thanks MMDetection team for the wonderful open source project!

If you find DeepInfrared useful in your research, please consider citing this project.

@article{dai2022oscar,
  title={One-Stage Cascade Refinement Networks for Infrared Small Target Detection},
  author={Yimian Dai and Xiang Li and Fei Zhou and Yulei Qian and Yaohong Chen and Jian Yang},
  journal={arXiv preprint},
  year={2022}
}