Optimization of Ship Small Target Detection Based on YOLOv10 in Complex Ocean Environment

Ship target detection technology plays a vital role in civilian maritime traffic monitoring and military maritime security protection, and it is the key to ensuring the safety and order of marine activities. In a complex ocean environment, the background information of optical images is complex. The ship images are missed and wrongly detected due to the change in UAV shooting height, so a CSCGhost target detection algorithm is proposed. Experiments denote that compared with the traditional YOLOv10, the mAP increases by 3.2%, the accuracy increases by 9.1 %, and the recall increases by 3.2%, which has a good detection effect and is especially suitable for offshore operations.