Article’s

Incidents involving wrong-way driving (WWD) represent a traffic danger often resulting in severe collisions and deaths. This article presents a detection system for wrong-way driving utilizing computer vision and deep learning techniques aimed at real-time traffic supervision and law enforcement. The approach integrates the YOLOv4 object detection model, centroid tracking for vehicle movement analysis and Automatic License Plate Recognition (ALPR), for identifying violators. Through observation of video feeds from traffic cameras the system accurately detects vehicles traveling against the flow captures their license plates and stores the data in a Firebase database for subsequent analysis and enforcement. This enhanced system incorporates improvements, over existing techniques, including refined centroid tracking methods, adaptive median line calibration and robust performance under challenging environmental factors. Test outcomes demonstrate that the system effectively and dependably identifies wrong-way driving events. The article provides an overview of the system design, algorithms, implementation specifics and experimental assessment highlighting the solutions capability to improve road safety and reduce WWD-related incidents.