Ship target detection based on YOLOv3 algorithm

As a coastal state, China's maritime trade and maritime security are crucial, and efficient and accurate detection and recognition of ships is an essential part of this. However, currently, deep learning-based ship target detection methods suffer from scarce datasets and imbalanced samples, leading to weak generalization capabilities of many detection models that are unable to adapt to a variety of environments. In this article, a new large marine ship dataset was established, and the AutoAugment data processing method was used to process ship image information. The YOLOv3 algorithm was utilized for ship target detection, improving the robustness and prediction accuracy of ship detection under complex coastal conditions, achieving high-quality detection of ships at sea.