Staged multi-target tracking and matching algorithm combined with depth estimation information

As the field of artificial intelligence progresses, multi-target tracking, one of the advanced visual tasks of computer vision, has developed rapidly. At present, as multi-target tracking technology gradually matures, related scientific research results are slowly appearing in our daily lives, bringing more powerful help to our lives. However, the precision of multi-target tracking faces significant constraints, mainly attributed to factors like occlusion and interference from the background. How to reduce the constraints of occlusion and other factors on the accuracy of multi-target tracking. In this paper, we propose a staged multi-target tracking and matching algorithm that combines depth estimation information. By using the depth estimation information of targets in the scene, combined with the proposed staged matching algorithm improves the precision of the algorithm for multi-target tracking.