Moving Object Detection and Tracking Technology Based on Hybrid Algorithm

Object tracking is a hot topic in visual technology and is widely applied in scenarios such as intelligent monitoring, autonomous driving, and robot visual perception.In recent years, with the sports industry's rapid development, tracking targets (balls and players) in complex sports scenes represented by basketball and football has attracted increasing attention.This paper focuses on tracking targets (balls as single targets and players as multiple targets) in competitive sports scenes like basketball and football.A small target detection network based on multi-scale features and a triangulation algorithm is employed to fuse the two-dimensional coordinates of the ball into three-dimensional coordinates.Additionally, a simplified motion model is proposed for the non-linear motion of the ball, and a Kalman filter is used to obtain accurate and smooth threedimensional tracking trajectories.The proposed method achieves 2D and 3D tracking accuracies of 0.81 and 0.92 on the basketball public dataset, respectively.