With the continuous advancement of urbanization and the sharp increase in the number of motor vehicles in cities, problems such as traffic congestion, frequent accidents, frequent violations and uneven distribution of road resources have become increasingly prominent, seriously affecting the efficiency of residents’ travel and the safety of urban operation. The traditional traffic management model mainly relies on centralized processing, which is carried out through traffic control system equipment, and manual intervention. The fixed model is difficult to meet the requirements of modern traffic systems for real-time performance, intelligence and scalability. To effectively enhance the intelligence level of urban traffic management, this paper proposes a lightweight deep learning method based on edge cloud computing, with YOLOv8n as the core detection model, and deployed in edge computing devices, to achieve an intelligent traffic detection system based on edge cloud collaborative computing. The experimental results show that the model detection frame rate on the edge computing device reaches 55.06 FPS, the F1 value reaches 0.84, and the recall rate on mAP@50 reaches 0.899.