International Journal of Scientific Research and Engineering Development

Road traffic violations such as helmet-rule violations, triple-riding, red-light jumping, and wrong-lane movement continue to increase the burden on traffic management agencies, particularly in densely populated urban regions. Manual monitoring of surveillance feeds is labor-intensive, difficult to scale, and often inconsistent under continuous operation. This manuscript presents an AI-powered traffic violation detection and evidence generation system designed to automatically analyze CCTV streams, identify rule violations in real time, and generate legally useful digital evidence records. The proposed framework integrates a deep-learning-based detector, multi-object tracking, automatic number plate recognition (ANPR), and rule-based event reasoning in a unified pipeline. The system is intended for deployment on both edge devices such as NVIDIA Jetson platforms and GPU-enabled traffic control servers. A major focus of the work is suitability for Indian road conditions, where dense traffic, regional license plate variations, inconsistent lane discipline, and mixed illumination create major challenges for visual analytics. The manuscript contributes a structured system architecture, formal event-scoring equations, deployment-oriented design choices, and a journal-ready methodological presentation aligned with the concept introduced in the project review presentation. The proposed model supports automated evidence packs consisting of image snapshots, violation labels, timestamp, recognized license plate text, and camera or location metadata. Such a framework can improve enforcement consistency, reduce dependence on manual observation, and support scalable smart-city traffic monitoring.

V.Sunil Anandh, Nithish Kumaar S, Mani Matheswaran M, Vignesh K, Sherine Samuel,"AI-Powered Traffic Violation Detection and Evidence Generation System" International Journal of Scientific Research and Engineering Development, V9(2): Page(1620-1630) Mar-Apr 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.