Abstract Infrared small target tracking serves as a critical technology for ensuring all-weather maritime security monitoring and operational stability in both military and civilian domains, yet faces dual challenges of redundant background interference and scarce target features that hinder conventional algorithms from achieving high-precision robust tracking in complex environments. Compounding these difficulties is the absence of large-scale datasets with precise annotations, which has significantly constrained technological advancements in infrared maritime small target tracking. To address these limitations, this study establishes TIR-SMALL-BOAT, the first dedicated infrared maritime small target tracking dataset comprising 17,352 precisely annotated infrared images across nine distinct maritime scenarios. Building upon this foundation, we propose the IMSTrack algorithm featuring an innovative Adaptive Scaling Field-of-View (ASFV) module specifically designed for infrared maritime small targets, which effectively suppresses background interference while maximally preserving critical feature information. The algorithm further enhances small target feature extraction through a parallel multi-scale spatial attention module and improves tracking precision using an optimized Power-IoU v2 loss function. Extensive experimental validation on our self-constructed TIR-SMALL-BOAT dataset demonstrates that IMSTrack outperforms mainstream trackers in both accuracy and stability.
Li et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: