基于YOLOv5的目标轮廓运动特征检测方法OA

The accurate recognition of target contour motion features is a critical technology in video analysis,particularly in fields of aerospace and manufacturing.Traditional feature detection methods often perform poorly in identifying and classifying targets,especially in multi-target environments and low-resolution challenges of small to medium-sized targets in videos.Therefore,this study proposes a target contour motion feature detection method based on YOLOv5.Utilizing YOLOv5,three-dimensional modeling of moving targets is achieved,and a staged approach is employed to extract target contour features.Ad-ditionally,convolutional neural networks are utilized to enhance feature learning,thus optimizing the detection process.The re-sults indicate that the proposed method achieves image frame detection at a speed of 29 frames per second with an average preci-sion of 91.74％,demonstrating its outstanding performance in detecting target contour motion features.