A HED-optimized Automatic Detection and Tracking Algorithm for Marine Moving Targets based on YOLO V3

Abstract

Target detection technology is one of the core technologies of modern military information technology research. It has important theoretical research and application value in the fields of military such as intelligence collection, key area monitoring and weapon guidance. At present, the traditional maritime target recognition method has poor timeliness, low precision, and poor self-processing capability. This paper proposes an autonomous detection and tracking algorithm for marine moving targets based on YOLO V3 algorithm optimized by HED. The method realizes the autonomous detection and tracking of marine moving targets through the optimization of YOLO V3 deep learning network and HED edge detection algorithm. This method has two advantages: (1) fast and accurate detection of maritime targets through the YOLO V3 deep learning network; (2) detection of target edges by HED edge detection algorithm to achieve the target and background segmentation, improving the accuracy of detection.