A fishing behavior detection method based on an improved YOLOv8 algorithm
In order to realize the intelligent and accurate identification of fishing behavior in irrigation district,an improved CM-YOLOv8 fishing behavior detection method was proposed. This method added a multi-scale feature fusion module (Conv-M)!to the backbone network of YOLOv8 to learn features from different Conv layers. At the same time,the self-learning weight coefficient was used to weight the features,so as to enhance the ability of the network to extract features from fishing behavior. The fishing behavior detection model was obtained by training the network,and then the fishing behavior in the video image data was detected and recognized. Experimental results on Fish-Data showed that compared with YOLOv8,the proposed method can increase precision by 1.1％,and recall by 1.4％,and mean average precision increased by 0.9％. The results could improve the intelligent level of supervision of fishing behavior in irrigation district.
国家科技期刊平台
NSTL
万方数据
