Traffic Monitoring System to Count Vehicles Driving Against the Flow of Traffic using Ultralytics Yolov8

Wrong-way driving poses significant risks to road safety and remains a critical challenge in traffic monitoring systems. This work addresses the problem by focusing on the detecting, tracking and counting of vehicles traveling against the designated traffic flow. An automated and reliable vehicle counting system is developed to support real-time traffic monitoring and incident detection. The system integrates the YOLOv8 object detection algorithm with the ByteTrack tracking method and is implemented within a Streamlit-based application to provide an intuitive interface for users. Two traffic scenarios, representing high-speed highway and low-speed urban conditions, are used to evaluate the system’s performance across multiple YOLOv8 model configurations. A Polygon-based Region Of Interest (PROI) approach is employed to enable accurate vehicle counting, ensuring adaptability to various camera angles and road layouts. Experimental results demonstrate low error rates and efficient operation in constrained computational environments, confirming the system’s potential for deployment in real-world traffic monitoring applications. This work highlights the effectiveness of computer vision techniques in enhancing road safety and traffic management, particularly in mitigating risks associated with wrong-way driving.