Queue length perception method based on one-way coupling between shockwave model and YOLOv5-DeepSORT
For the real-time perception of queue lengths at intersections,a method was proposed that combined traffic mathematical models with intelligent perception devices for queue length detection.The method determined the maximum queue length of the road using shock wave models,which served as input for a video perception model based on YOLOv5-DeepSORT,to achieve unidirectional coupling between traffic mathematical models and intelligent perception models.To validate the effectiveness and superiority of this method,real-time queue length perception was conducted at a specific intersection in Lanzhou City.Perception data from different time periods at the selected intersection were investigated to explore the impact of varying traffic flow on the perception accuracy of this method.The research findings indicated that the queue length detection area determined by the queue length perception method based on shock wave models and YOLOv5-DeepSORT unidirectional coupling outperformed the control group in overall perception accuracy.Significant improvements were observed in metrics such as mean absolute error,root mean square error,and mean absolute percentage error,with an increase in accuracy of over 40%in certain scenarios.
万方数据
维普
