Advanced classified turning traffic volume module at Unsignalized Intersection

Intersection study is one of the crucial surveys for advanced traffic management, especially at unsignalized points. In this study, the traditional method for vehicle detection and classification is initially implemented to classify traffic volume at unsignalized intersections. To enhance the accuracy of object detection and counting, a deep learning approach for vehicle classification and counting is designed using the YOLO (You Only Look Once) algorithm and a modified YOLO algorithm. This includes a variation of a statistical approach for the best outlier detection. The approach aims to improve the efficiency of vehicle classification and counting at unsignalized intersections, which is a critical task in intelligent transportation systems. The results verified for four classes of vehicles and indicate that the deep learning approach with an effective outlier detection approach increased the efficiency of vehicle classification and counting by 30 to 40 percent with satisfactory results.