Study on Marine Fishery Law Enforcement Inspection System based on Improved YOLO V5 with UAV

As a part of China’s territory, the wide sea not only represents China’s powerful territorial power, but also contains incomparably rich Marine resources. And the vast area of the sea has led to a lack of inadequate maritime supervision. The abundant fishery resources are one of the important economic sources of China, but the law enforcement of ocean and fishery has long been plagued by the contradictions among the wide sea area, long coastline, large number of fishing boats in fishing ports, difficult management and limited law enforcement. It is difficult to achieve full-time, all-directional and all-coverage control. Relying on the existing Marine and fishery law enforcement personnel, vehicles and vessels, equipment and so on, it is difficult to achieve sufficient and effective law enforcement supervision. At present, the combination of UAV technology and target detection has a wide application prospect, providing a new technical means for Marine and fishery law enforcement. In this paper, the improved YOLO V5 algorithm is configured on the Unmanned Aerial Vehicle (UAV) to further improve the work efficiency of the UAV. Compared with the initial YOLO V5, the paper changed the 3-scale feature extraction of Neck end into 4-scale feature extraction. By integrating more shallow feature information of more scales, the feature expression ability of path aggregation network was enhanced, the detection accuracy of small targets was improved, and the missed detection rate was reduced. At the same time, the improvement of boundary box detection function and non-maximum suppression NMS improves the research of target recognition and target overlapping detection. Finally, the training results show that the improved YOLO V5 can further improve the effectiveness of UAV Marine fishery law enforcement.