FVSI-YOLO: A Full-View Ship Identification and Tracking Method for Drones

This paper presents FVSI (Full-View Ship Identification)-YOLO, a comprehensive sea vessel identification and tracking method tailored for drone applications, based on the YOLO framework. By utilizing altitude information from drones, this method effectively distributes identification tasks between high-altitude overhead views and low-altitude horizontal views. The model incorporates the latest YOLOv8 architecture with Oriented Bounding Box (OBB) rotation technology, enhancing the capability to identify vessels from overhead angles. It is particularly effective for dynamic sea surface background target detection tasks. Experimental results demonstrate the model’s superior speed, accuracy, stability, and reliability across diverse marine environments, underscoring its significant potential for maritime vessel monitoring applications.