Ship Detection: An Improved YOLOv3 Method

Abstract

YOLOv3 is the state of art detector, which performs an excellent balance in detection speed and accuracy. In this paper, an improved YOLOv3 model named YOLOv3-ship is proposed for the ship detection. The main contributions to the YOLOv3-ship consists of dimension Clusters, network Improvement and embedding of the Squeeze-and-Excitation(SE) module. The experiments results show that the detection accuracy has been significantly improved by the YOLOv3-ship.