Real-time traffic management system based on YOLOv8 and DeepSORT in a complex traffic environment

With increasing urban traffic complexity and vehicle growth, traditional management systems face significant challenges. This paper introduces an intelligent traffic management system based on the You Only Look Once v8 detection model and DeepSORT tracking algorithm, integrating modules for pedestrian and vehicle detection, license plate recognition, vehicle speed measurement, distance estimation, and traffic light recognition. A user-friendly GUI was developed with PyQt5 to support image, video, and real-time camera input, enabling real-time tracking, speed, and distance estimations. Each module’s design and implementation are detailed, and tests in complex traffic environments have shown high detection accuracy and system stability. Compared to a baseline model, this system demonstrates improved performance across traffic monitoring tasks, providing a robust and cost-effective approach to modern traffic management.