Improved YOLOv5s algorithm for small item detection of wheelhouse

Abstract

With the development of deep learning, object detection has achieved rapid development in recent years, and is widely used in real-life scenarios such as face detection and automatic driving. In the field of ship navigation safety, it is necessary to identify whether some specific items appear in the wheelhouse to help determine whether there is a threat to drive safety. These items are usually small in size and require higher detection efficiency. To address this problem, this paper proposes a ship-specific item detection method that improves the YOLOv5s algorithm. By introducing the convolution attention mechanism module CBAM, the feature extraction ability of the network, the detection capability of small targets, and the detection accuracy are improved. The experimental results show that after the introduction of the attention mechanism, the precision rate of YOLOv5s on ship-specific items is 85.6%, the recall rate is 85.2%, and the average accuracy is 90.2%, which can complete the detection task of specific items of wheelhouse small targets