A Sonar Image Target Detection Method with Low False Alarm Rate Based on Self-Trained YOLO11 Model

Autonomous detection of sonar image targets is a key technology for unmanned undersea systems, but it faces the challenge of high false alarm rates in practical applications, which limits the quality and efficiency of mission execution by unmanned underwater systems. In this paper, an underwater target detection method based on the YOLO11 model was designed, and a false alarm rate detection method by self-training a deep learning detector on sonar images was proposed to reduce the false alarm rate. This method automatically generated proxy classification tasks based on the sonar image target detection dataset and improved the deep learning detector’s learning of target and background features through pre-training, enhancing the detector’s ability to distinguish between targets and backgrounds and thereby reducing the false alarm rate. Experimental results demonstrate that when the detector’s confidence is set to the value corresponding to the maximum F1-score, the YOLO11 detector trained using the proposed method can reduce the false alarm rate by 11.60% compared to traditional transfer learning methods while achieving a higher recall rate. This method improves the generalization of the deep learning detector without using external datasets, providing an efficient self-training approach for underwater target detection scenarios with small sample sizes.