基于无人机视频的交叉口全景感知技术研究OA

Urban intersections,as common bottleneck nodes in road traffic,play a crucial role in regulating,organizing,and controlling traffic flow.Utilizing modern information technology to comprehensively perceive the traffic operation status at intersections and achieving the intelligent management of urban traffic systems is an important research topic in current traffic management.This paper proposes a panoramic perception method for intersections based on low-altitude drone video footage.By constructing a traffic target detection and tracking model using the YOLOv5 and Deep SORT algorithms,this method accurately identifies static and dynamic traffic elements within the video range and intelligently extracts the global trajectory information of dynamic traffic elements at the intersection,providing data support for refined intelligent traffic control at intersections.The method was verified at two road intersections in Zhuhai City.The results show that the accuracy of vehicle detection and tracking results is above 99.8%.This method is characterized by reliability,convenience,and broad applicability,offering a new perception and analysis model for traffic control in the context of big data.