ASSESSING ROAD SAFETY AT INTERSECTIONS USING COMPUTER VISION AND CRASH DATA ANALYTICS

Abstract

Intersections form vital nodes in urban traffic networks and are collision-prone areas due to intricate vehicle-pedestrian interactions. Road safety is studied at intersections employing an integrated approach consisting of computer vision and crash data analytics. Having video images from traffic surveillance cameras, the computer vision algorithm identifies and tracks road users, studies vehicle trajectories, and identifies near-misses such as hard braking, red-light violations, and unsafe lane changes. Meanwhile, historical crash data are studied to identify patterns that are highly frequent and severe at various types of intersections. By correlating behavioral indicators from video analysis with long-term crash statistics, the framework forms a synergistic view of safety risk and potential causes for accidents. These findings provide a fertile ground for thinking about proactive safety interventions, including better signal timing, road design changes, and enforcement strategies that are data-driven. Hence, this interdisciplinary approach enriches traffic safety assessment and smart urban planning.