Dangerous driving on the highway is a significant safety hazard, and existing highway cameras are not used to detect whether drivers are fatigued or distracted. To inform the police of dangerous driving vehicles promptly,In this paper, we propose a complex driving detection system on the basis of YOLO network.The YOLOv8s algorithm is used for real-time analysis, effectively improving detection accuracy and caters to the need for lightweight design. The system combines the characteristics of highway camera settings with distributed cluster deployment, solving the defects of detection blindness and target loss and achieving the function of distributed clustering to detect drivers of passing vehicles. The system can provide adequate support for traffic police to intercept dangerous drivers on highways and give theoretical guidance for the future intelligent transformation of road traffic through experimental verification.The technology could also be rolled out to cover urban roads and additional transportation networks to further ensure traffic safety.