Video-Based Traffic Flow Analysis for Turning Volume Estimation at Signalized Intersections

Traffic flow analysis in complex areas (e.g., intersections and roundabouts) plays an important part in the development of intelligent transportation systems. Among several methods for analyzing traffic flow, image and video processing has emerged as a potential approach to extract the movements of vehicles in urban areas. In this regard, this study develops a traffic flow analysis method, which focuses on extracting traffic information based on Video Surveillance (CCTV) for turning volume estimation at complex intersections, using advanced computer vision technologies. Specifically, state-of-the-art techniques such as Yolo and DeepSORT for the detection, tracking, and counting of vehicles have enveloped to estimate the road traffic density. Regarding the experiment, we collected data from CCTV in an urban area during one day to evaluate our method. The evaluation shows the proposing results in terms of detecting, tracking and counting vehicles with monocular videos.