Ship Detection Algorithm based on Improved YOLO V5

With the rapid development of computer vision and machine vision, deep learning based methods have achieved good results in the field of target detection, recognition, and tracking. However, for ship detection and recognition on the sea surface, the detection and recognition rate is greatly affected by the uneven distribution of the horizontal and vertical features for ships and the different sizes of ships. In order to improve the ship detection accuracy and real-time performance, this paper proposed a ship detection algorithm based on YOLO V5, in which the feature extraction process was merged with the GhostbottleNet algorithm. Specifically, the algorithm consisted of two stacked GhostNet to refine and capture the image features, so as to overcome the incomprehensive feature capture problem in the original YOLO V5 network due to the inhomogeneous distribution of ship image features in transverse and vertical. Experimental results show that the proposed method not only improves the detection accuracy of YOLO V5 algorithm but also makes the GIoU decrease steadily.