Lightweight ship detection method based on YOLO-FNC model

A lightweight and efficient ship detection method based on YOLO-FNC was proposed for the complex environment such as the port with dense traffic. First, a neural network module FasterNext based on the FasterNet method is designed, and this module replaces the C3 module in the YOLO model to ensure faster operation without affecting the accuracy. Second, the NAM(Normalization-based Attention Module) is integrated into the network structure, and the sparse weight penalty is used to suppress the feature weights to ensure more efficient weight calculation. Finally, a new bounding box regression loss is proposed to speed up the prediction frame adjustment and increase the regression rate to improve the rate of convergence of the network mode. The experimental results show that the proposed method performs detection experiments on ship datasets in a self built complex scene, in which improve mAP@0.5 by 6.3%, reduce parameter count by 9.74%, and reduce computational complexity by 11.4%, effectively achieving lightweight and high-precision ship detection compared with the YOLOv5s algorithm.