Search papers, labs, and topics across Lattice.
This paper tackles the challenge of cross-domain infrared small target detection (IRSTD) where models trained on one domain fail to generalize to new observational conditions. They identify inconsistencies in spectral phase as a key factor causing domain discrepancies and propose a Spatial-Spectral Collaborative Perception Network (S$^2$CPNet) to address this. S$^2$CPNet incorporates a phase rectification module (PRM), orthogonal attention mechanism (OAM), and selective style recomposition (SSR) to derive generalizable target awareness and mitigate domain-specific biases, achieving state-of-the-art performance across diverse cross-domain IRSTD datasets.
Domain shift in infrared target detection is primarily a spectral phase problem, and can be largely solved with a frequency-aware network.
The accurate target-background separation in infrared small target detection (IRSTD) highly depends on the discriminability of extracted representations. However, most existing methods are confined to domain-consistent settings, while overlooking whether such discriminability can generalize to unseen domains. In practice, distribution shifts between training and testing data are inevitable due to variations in observational conditions and environmental factors. Meanwhile, the intrinsic indistinctiveness of infrared small targets aggravates overfitting to domain-specific patterns. Consequently, the detection performance of models trained on source domains can be severely degraded when deployed in unseen domains. To address this challenge, we propose a spatial-spectral collaborative perception network (S$^2$CPNet) for cross-domain IRSTD. Moving beyond conventional spatial learning pipelines, we rethink IRSTD representations from a frequency perspective and reveal inconsistencies in spectral phase as the primary manifestation of domain discrepancies. Based on this insight, we develop a phase rectification module (PRM) to derive generalizable target awareness. Then, we employ an orthogonal attention mechanism (OAM) in skip connections to preserve positional information while refining informative representations. Moreover, the bias toward domain-specific patterns is further mitigated through selective style recomposition (SSR). Extensive experiments have been conducted on three IRSTD datasets, and the proposed method consistently achieves state-of-the-art performance under diverse cross-domain settings.
