From Weak to Strong: Cross-Level Energy-Field Aggregation for Infrared Small Target Detection

Infrared small target detection (IRSTD) remains a challenging task due to the lack of discriminative texture cues in targets and the presence of highly complex backgrounds. Inspired by the spectral energy distribution characteristics inherent to infrared imagery, this paper develops a novel Cross-layer Energy Field Aggregation Network (CEFANet), which focuses on target energy discontinuities embedded in cluttered backgrounds to achieve dynamic discrimination and feature enhancement of small targets with varying states. Specifically, to effectively exploit the fact that infrared targets typically manifest as high-energy responses at local scales, we introduce an Energy Field Attention (EFA) module that selectively amplifies target energy mutations while suppressing background noise. In addition, we propose a Cross-layer State Fusion (CSF) module based on channel-wise state awareness and adaptive weighting mechanisms, enabling precise modulation of low-level features under the guidance of high-level semantic states. Extensive experiments conducted on multiple public benchmarks demonstrate that the proposed CEFANet significantly improves the detection performance of infrared small targets.