Ship target detection in complex background b...

Ship target detection in complex background based on Yolo network

Author Name : Zhang Huahui, Sun Dechao, Huang Hong, Cheng Zexian

ABSTRACT

Ship target detection in inland waterway is of great significance for port management and other maritime inspection tasks. Due to the low illumination environment such as fog and night and the interference of small target ships, the traditional target detection method is difficult to achieve effective ship target recognition. Therefore, this paper proposes an improved method based on YOLO algorithm for the complex river environment. By introducing SPP structure, the receptive field in the process of ship feature extraction is enhanced, and the expression ability of multi-scale feature map is enhanced. The experimental results show that the algorithm achieves good recognition effect in night and fog scenes, and improves the phenomenon of small target ship missing detection.

Keywords: Ship target detection, SPP structure, YOLO algorithm.