Design and Implementation of Pedestrian Target Tracking System Based on YOLO Architecture

Pedestrian tracking is an essential issue in the field of computer vision. Pedestrian tracking can be applied to safety monitoring in public places, such as airports, stations, shopping malls, etc. It can monitor the location and action trajectory of pedestrians in public places in real-time, which can quickly detect abnormal behaviors and take timely measures to prevent lawless elements from using public places for criminal activities, thus ensuring public safety. With the development of deep learning, the performance of pedestrian tracking algorithms is gradually improving. This article designs and implements a pedestrian target tracking system based on YOLOv5 and DeepSORT. The system introduces an excellent YOLOv5 detector into the DeepSORT algorithm, detects the tracking video frame by frame, and uses Kalman filtering to predict the target position. At the same time, the Hungarian algorithm is used to match the same target to achieve pedestrian tracking. After testing, this algorithm can significantly improve the issues of identity transformation and tracking errors caused by occlusion. At the same time, using PyQt5 to develop the system interaction interface, the system interface is straightforward, easy to understand and use. This system can help users better understand pedestrian movements, quickly respond to possible abnormal behaviors, and ensure social safety.