Panoramic Traffic Monitoring Target Detection And Tracking And Its Application In Embedded System

In the urban intelligent transportation system,panoramic traffic monitoring target detection and tracking has become a research hotspot.The technology of target detection and tracking based on deep learning has developed more and more rapidly,which is widely used in the field of urban intelligent transportation.Through the detection and tracking of the target in the monitoring video,the target positioning information and track information are obtained,analyzed and intelligentized,and the parameter information of the traffic target is obtained to analyze the traffic situation,formulate reasonable traffic rules and early warning measures,and create intelligent traffic.In view of the characteristics of complex urban traffic scene,multiple vehicles and pedestrians and large scale changes,an improved multi-target detection method for panoramic traffic monitoring of YOLOv3 is proposed.Firstly,based on the YOLOv3 network,four detection scales are designed considering the characteristics of large and small scale targets,and four detection scale features are fused.Then,K-means clustering method is used to cluster the labeled target boxes in the dataset,and the optimized width and height dimensions of clustering anchor boxes are selected as the initial candidate boxes of the improved YOLOv3 network.Panoramic traffic monitoring and detection targets include five categories:large cars,small cars,motorcycles,bicycles and pedestrians.On the test set,the average accuracy and recall rate of target detection are 84.49% and 97.18% respectively,which are7.76% and 4.89% higher than the original YOLOv3.The processing speed can meet the requirements of real-time detection in traffic scenes.The improved YOLOv3 detection model is transplanted into the embedded GPU platform for application.The video target detection system and the static image target detection system are designed respectively to detect the urban traffic video target and the monitoring image target,and the number of detected traffic targets is counted to obtain the traffic information parameters.This paper analyzes the operation of video and surveillance image targets.After the experimental test,the video and still image detection system has high detection performance and certain application value.Finally,based on the combination of detection and tracking,the optical flow tracking algorithm is used to track the panoramictraffic monitoring target,and the tracking accuracy is more than 90%.And it can track the specific target,print the tracking track,and get the target running track information,which is convenient for the traffic department to monitor effectively.