The evolution of maritime surveillance is sig-nificantly marked by the incorporation of Artificial In-telligence and machine learning into Unmanned SurfaceVehicles (USVs). This paper presents a an AI methodfor detecting and tracking unmanned surface vehicles,specifically leveraging an enhanced version of YOLOv8,fine-tuned for maritime surveillance needs. Deployedon the NVIDIA Jetson TX2 platform, the system fea-tures an innovative architecture and perception mod-ule optimized for real-time operations and energy effi-ciency. Demonstrating superior detection accuracy witha mean Average Precision (mAP) of 0.99 and achievingan operational speed of 17.99 FPS, all while maintain-ing energy consumption at just 5.61 joules. The remark-able balance between accuracy, processing speed, and energy efficiency underscores the potential of this sys-tem to significantly advance maritime safety, security,and environmental monitoring.