This work introduces a real-time traffic violation detection and automated enforcement system using Computer vision methodologies, notably YOLOv5 and Mask R-CNN. The system seeks to enhance traffic management by the automated detection and classification of traffic infractions, including red light violations, helmet non-compliance, mobile phone use, speed violations, and lane departures. YOLOv5 is used for rapid and precise object recognition, whilst Mask R-CNN is implemented for detailed segmentation and violation classification. The identified infractions are validated using a probabilistic method grounded in Bayesian reasoning, guaranteeing minimally false positive rates. Upon confirmation of a violation, the system initiates automatic enforcement measures, including the imposition of penalties or notification of authorities, therefore minimizing human involvement in traffic law enforcement. The system incorporates real-time data recording, preserving infraction records for further analysis and reporting. This study illustrates the capability of integrating AI-driven object identification with automated enforcement to develop intelligent, efficient, and scalable traffic monitoring systems, therefore enhancing road safety and improving traffic control. This study uses 'Traffic Violation Detection Dataset' given by Roboflow. It comprises annotated images of diverse traffic infractions, including red light violations, helmet noncompliance, and mobile device use, facilitating precise model training. It highlights the ability to transform traffic management and enforcement practices in smart cities by integrating computer vision and IoT technologies. © 2025 Elsevier B.V., All rights reserved.