REAL-TIME VEHICLE TRACKING AND COUNTING USING YOLOV8

Abstract

Vietnam is a developing country with a considerable demand for commodity transportation, so the number of vehicles that move on the street every day is huge. As a result, the number of traffic accidents also increased significantly. Therefore, an efficient management system is now required due to the large rise in the number of vehicles on the road as a result of this. To help solve this problem, a computer software that can count, classify, and detect automobiles from an image or video source using combined technologies such as YOLOv8, OpenCV, and CustomTkinter to build the GUI is needed. Vehicles, including cars, trucks, motorbikes, and buses, that are frequently seen on the road can all be counted using the software. The YOLOv8 custom model gained impressive results, showcasing an average accuracy of 0.94 (mAP@.50) and 0.747 (mAP@.50-95) on the test set across four classes. These findings underscore the model's robust performance in object detection. Upon identification of vehicles in images and videos, the model provides valuable insights by disclosing both the number and type of vehicles present. This information serves as a powerful tool for traffic managers, enabling them to access more intuitive data that facilitates efficient traffic management and strategic planning. The model's ability to deliver detailed insights enhances the overall effectiveness of traffic control measures.