I. Introduction
Counting the number of vehicles' turning movements in traffic video sequences has become a vital component of modern Intelligent Transportation Systems (ITS). By monitoring the flow of vehicles, crucial information can be obtained, including traffic status, lane occupancy, and congestion levels. This data can be used to provide early warnings of potential accidents, prevent congestion, and enhance automatic navigation. Moreover, by analyzing traffic flow information at various intersections and at different times, it is possible to optimize dynamic signal timing, leading to improved traffic flow and significant economic benefits. In the field of Intelligent Transportation Systems (ITS), traditional methods for counting vehicles have relied on specialized sensors such as magnetic coils[1], microwave[2], or ultrasonic[3] detectors. While these methods have provided some level of accuracy, they also have limitations in obtaining detailed information and are often expensive to install[4].