The increase in rapid urbanization worldwide has made traffic congestion worse, even as traditional fixed-time traffic systems poorly respond to changes in traffic. The problem is exacerbated through the delay of vehicles moving through an intersection, increasing the delay, fuel consumption, and pollution. In this paper we will implement a full-scope traffic control system using state-of-the-art artificial intelligence methods in our case using computer vision and machine learning to provide a dynamic, adaptive response. The implementation is based on collecting real-time data from surveillance cameras at the intersections. There is a vehicle identification using the YOLOv3 object-detection model, supporting an accurate identification and classification of vehicles, while generating a continued stream of traffic metrics (i.e. traffic density and vehicle flow). Inputs to predictive traffic control algorithms based MX traffic control using machine-learning will adjust signal timings to support more efficient traffic flow, decreasing delays and queues in the intersection. We implement the system using a simulated study with real traffic data and report on both average wait time and queue lengths at the intersections. The system was found to outperform the traditional traffic signal systems resulting in significant delays with traditional traffic control systems and improvement traffic management. The present congestion fundamentally arises from the increasing amount of vehicles that are both depleting and contaminating the environment. Consequently, a traffic control system is a major component of smart city agendas that attempts to augment urban mobility and safety as well as sustainability. Because traditional traffic control systems use fixed timing plans and/or limited sensor algorithm systems, the proposed research will first discuss the background on and overview traditional traffic signal control systems; then this review will illustrate a way forward to future plans related to smart traffic signal systems. Ultimately, the goal is to present the effectiveness of in the adapting algorithm of AI for improving urban traffic management and mobility.