Design and Implementation of a Wrong Side Vehicle Detection System Using YOLO And OpenCV.

Abstract: The wrong-side driving among the most significant contributing factors to serious traffic crashes, congestion, and loss of life in urban and highway transportation systems. The traditional approaches to traffic monitoring are mostly manual, time consuming and subject to human error. This paper proposes an AI based Wrong Side Vehicle Detection System (WSVDS) to automatically detect vehicles that travel in the opposite direction of the traffic flow using CCTV surveillance camera by computer vision and deep learning techniques. The proposed system relies on the YOLO (You Only Look Once) algorithm to detect vehicles in a frame and Deep SORT/Centroid Tracking to track the vehicles' movement between the frames. Direction analysis is carried out by comparing the trajectory movements of the vehicles with the previously defined traffic direction. When a vehicle travels in the opposite direction to the authorized direction, the system recognizes this as an incident and produces alerts and the storage of evidence. The system can accurately and quickly identify cars, bicycles, buses and trucks in real time, and has a low identification latency. The experimental results show an accuracy of around 95%, which shows the suitability of the proposed system for smart city surveillance and intelligent transportation systems. The proposed solution will greatly cut down on the need for man-to-man monitoring and enhance the safety of the road through automated traffic violation detection.

Keywords: Wrong Side Detection, YOLO, Deep Learning, Computer Vision, Traffic Surveillance, Intelligent Transportation System, Vehicle Tracking, CCTV Monitoring, Deep SORT, AI-based Traffic Monitoring.