Beyond Full Labels: Energy-Double-Guided Single-Point Prompt for Infrared Small Target Label Generation
Shuai Yuan, Hanlin Qin, Renke Kou, Xiang Yan, Z. Li, Chenxu Peng, Dongliang Wu, Huixin Zhou
Abstract
In this article, we pioneer a learning-based single-point prompt paradigm for infrared small target label generation (IRSTLG) to lobber annotation burdens. Unlike previous clustering-based methods, our intuition is that point-guided mask generation just requires one more prompt than target detection, i.e., IRSTLG can be treated as an infrared small target detection (IRSTD) with the location hint. Therefore, we propose a simple yet effective energy-double-guided single-point prompt (EDGSP) framework, aiming to adeptly transform a coarse IRSTD network into a refined label generation method. Specifically, EDGSP comprises three key modules: first, target energy initialization, which establishes a foundational outline to streamline the mapping process for effective shape evolution, second, double prompt embedding for rapidly localizing interesting regions and reinforcing high-resolution individual edges to avoid label adhesion, and third, bounding box-based matching for eliminating false masks via considering comprehensive cluster boundary conditions to obtain a reliable output. In this way, pseudolabels generated by three backbones equipped with our EDGSP achieve 100% object-level probability of detection (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">${{P}_{d}}$</tex-math></inline-formula>) and 0% false-alarm rate (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">${{F}_{a}}$</tex-math></inline-formula>) on SIRST, NUDT-SIRST, and IRSTD-1k datasets, with a pixel-level intersection over union improvement of 13.28% over state-of-the-art label generation methods. Further applying our inferred masks to train detection models, EDGSP, for the first time, enables a single-point-generated pseudomask to surpass the manual labels. Even with coarse single-point annotations, it still achieves 99.5% performance of full labeling.