EH-YOLO: Ship Target Detection Network Based on Improved YOLOv8

With the development of global economic integration, shipping traffic is becoming increasingly busy and the risk of traffic accidents has increased. Visual sensors, due to their wide field of view, low cost, and wide applicability, can provide richer information, which helps to make timely and accurate decisions, thus ensuring the safety of navigation. Deep learning-based object detection algorithms with excellent detection rates and efficient speeds have been widely used in various scenarios. This study aims to improve the YOLOv8 model to enhance ship detection capability. In this paper, the EffectiveSE attention mechanism is added to the C2F part of the YOLOv8 framework, and the Halo Attention self-attention mechanism is used to replace the detection head. By using Shape-IoU loss function and PReLU activation function, the accuracy of bounding box localization and the convergence speed of the model are improved. The proposed EH-YOLOv8 model shows excellent detection performance in complex scenarios, which helps to improve the safety of ship navigation and regulatory efficiency