The swift urbanization has considerably led to an increase in the number of vehicles, rendering the traditional fixed- time traffic light systems less effective. These systems are unable to respond to the changing traffic conditions in real-time and they also lack the capacity to give priority to emergency vehicles, often resulting in traffic jams and delays. The paper proposes a smart traffic control system that leverages computer vision and AI to manage traffic lights in a more efficient way as per the density of vehicles. The model utilizes video (both live and recorded) for detecting and counting different vehicles on multi- lane roads through a vehicle detection model based on YOLO. In the context of traffic-management, a dynamic-control algorithm which accords traffic signal timing according to traffic demand and a better-priority override for ambulances are some of the features of the system. Implemented through Python and Flask, the system allows real-time observation, simulation, and data analysis via a web-based interface. The experiments have shown a more efficient traffic-flow, lesser waiting time, and proper emergency response. The proposed method is however a scalable and affordable way to manage traffic in smart cities. Besides on-the-go traffic management, the system also archives traffic data for reviewing and plotting, thus allowing to comprehend the traffic flows and identifying the times of the highest congestion. The modular structure of the system guarantees the ability to be improved later on by features like multi-intersection coordination and predictive traffic management. Therefore, the proposed system is more than just a real-time solution; it also serves as a base for data-driven urban traffic optimization.
