Consistent Learning of Sparse Background Features for Infrared Small-Target Labeling

Hai Xu, Hui Zhou, Sheng Zhong, Luxin Yan, Xu Zou

Abstract: Recent years have witnessed many remarkable achievements in infrared small-target (IST) detection based on deep learning (DL). To achieve high performance in real-world applications, DL methods require a substantial number of accurate labels. However, IST annotating is labor-intensive as they are very small. Determining and annotating edge pixels demands considerable time and effort, which slows down data expansion and further research in IST detection. To mitigate the issue, we propose a pseudo-label (PL) generation method named consistent learning of sparse background feature (CLSBF). This approach models the generation of IST PLs as a domain transformation from local target maps to background ones. It can relax the supervision requirements of DL methods, transitioning from absolute pixel-level supervision to point supervision. This method employs an unsupervised approach primarily, supplemented by semi-simulated supervision, to achieve mutual conversion of local images from different domains and obtain the final target PLs through the differences between target images and transformed background ones. Experiments show that equipped with our method, models trained with the input of a coarsely accurate center label can achieve performance up to 99.94% compared to models trained with official accurate labels. Furthermore, when the official labels are not accurate enough, models trained with PLs generated by CLSBF consistently show a performance improvement of 1.11%–4.09% when they are evaluated based on re-labeled bounding boxes. Extensive experiments demonstrate the reliability of CLSBF in generating PLs and its potential to alleviate the labor-intensive process of manual labeling significantly. We have released an IST labeling tool with CLSBF as an assistant at https://github.com/SeaHifly/CLSBF_software.git