HyTBE: Hyperbolic Model for Cross-Domain Infrared Small Target Detection

A new research paper titled 'HyTBE: Hyperbolic Target-Background Expert Model for Cross-Domain Infrared Small Target Detection' has been published on arXiv (ID: 2608.05771). The paper addresses the problem of performance degradation in infrared small target detection (IRSTD) when models are applied to unseen infrared domains. Existing methods focus on enhancing target responses and suppressing background interference, but they often fail when trained on limited source domains because the learned decision rules are based on restricted target-background relation patterns. The authors introduce HyTBE, a Hyperbolic Target-Background Expert model, which aims to expand source-domain relation patterns and adaptively adjust visual representations using explicit... (the abstract is cut off). The paper is categorized as a cross-domain study and was announced as a cross-type submission. The research is relevant to the fields of computer vision and machine learning, particularly for applications in surveillance and remote sensing where infrared small target detection is critical.