Research on Detection and Tracking Algorithm for Small and Weak Targets on the Sea Surface Based on Neural Networks

To address practical issues such as complex backgrounds, indistinct target features, and poor real-time performance in detecting and tracking small and weak targets at sea, this paper first designs a guided filter algorithm fused with a foggy day imaging model for image preprocessing. This aims to simplify the background complexity. Subsequently, the incremental YOLOv8 detection head algorithm is employed for detecting and tracking these targets, while integrating the ACmix hybrid model to enhance overall operational efficiency. Finally, simulation experiments are conducted to validate the effectiveness and feasibility of the proposed algorithm.