Abstract :

This paper presents a computer-vision approach for lane detection and overtaking behavior analysis using roadside or dashboard camera video. The proposed system detects lane boundaries in real time, identifies vehicles using object detection and tracking, and analyzes overtaking maneuvers based on spatial and temporal criteria. A robust lane detection pipeline integrated with vehicle classification and motion tracking enables accurate detection of overtaking events in complex traffic scenes. Experimental results on real traffic video demonstrate high accuracy in lane boundary localization and successful overtaking detection under various weather and lighting conditions. The system is suitable for intelligent transportation applications like Advanced Driver Assistance Systems (ADAS) and traffic surveillance.