Research on Pedestrian Object Detection and Trajectory Tracking Methods

Pedestrian multi-target tracking is a popular direction of computer vision, which is applied to intelligent surveillance, traffic, unmanned driving and other fields, and has broad prospects. However, due to the complex monitoring environment, pedestrian occlusion and variable movement, there are still many problems in this technology. Based on the requirements of intelligent monitoring, this paper studies a real-time and effective detection and tracking algorithm, and uses YOLOv5 and DeepSORT algorithms to realize pedestrian detection and trajectory tracking. At the same time, the trajectory tracking and crowd density statistics functions are added to improve the monitoring effect.