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

2026-08-06

AI Summary

This work addresses the performance degradation of infrared small target detection in cross-domain scenarios, primarily caused by shifts in target-background relational patterns. To mitigate this issue, the paper proposes HyTBE, a novel approach that uniquely integrates hyperbolic space modeling with a Mixture-of-Experts (MoE) mechanism. HyTBE explicitly intervenes in target-background relationships during training to enrich relational pattern diversity, embeds multi-scale features along with target and background anchors as relative distances in the Poincaré ball, and employs a hyperbolic-guided MoE adapter to dynamically calibrate features. Leave-one-domain-out experiments on NUAA-SIRST, NUDT-SIRST, and IRSTD-1K demonstrate that HyTBE significantly outperforms existing methods, effectively enhancing cross-domain generalization capability.

Abstract

Infrared small target detection (IRSTD) has achieved substantial progress under domain-consistent evaluation, yet detector performance often degrades markedly when generalizing to unseen infrared domains. Existing methods primarily improve detection by enhancing target responses and suppressing background interference. However, when trained on only a limited set of source domains, their learned decision rules are inevitably established from a restricted range of source-domain target-background relation patterns. We formulate this cross-domain failure as target-background relation shift: unseen domains may exhibit relation patterns that are not observed during training, thereby weakening the discriminative capability learned from the source domains. To address this problem, we propose HyTBE, a Hyperbolic Target-Background Expert model that expands source-domain relation patterns and adaptively adjusts visual representations using explicit relation cues. The Target-Background Relation Intervention selectively perturbs either targets or backgrounds, broadening the observable relation patterns during training while maintaining valid supervision. Subsequently, the Hyperbolic Relation Modeling maps multi-scale visual cues into a Poincaré ball and characterizes the target-background relation of each feature token according to its relative distances to the target and background anchors. The Hyperbolic-guided MoE Adapter further uses these hyperbolic relation representations to calibrate multi-scale visual features and aggregate expert-specific feature corrections for different relation patterns. Leave-one-domain-out experiments on NUAA-SIRST, NUDT-SIRST, and IRSTD-1K demonstrate that HyTBE achieves stronger cross-domain generalization than competitive baselines.

Problem

• cross-domain generalization
• infrared small target detection
• target-background relation shift
• domain shift
• IRSTD

Innovation

• Hyperbolic Geometry
• Target-Background Relation
• Cross-Domain Generalization
• Mixture of Experts
• Infrared Small Target Detection