The smart city concept has aroused interest around the world, including governments, businesses, universities, and institutes. A smart city can utilize the potential of information technology (IT) to promote sustainable development and improve the quality of life of its citizens more efficiently. Vehicular traffic in large cities represents a major problem, as it affects various aspects of urban life. These problems can include stress on people, late arrival at their destinations, increased risk of road accidents, and environmental pollution, among others. The use of Internet of Things (IoT) solutions in the real world has increased exponentially. In smart cities, networked IoT devices are collecting data from the physical environment to optimize decisions to improve city services to citizens. This paper introduces an algorithm designed to alleviate traffic congestion in large cities. It utilizes a You Only Look Once (YOLO) model to precisely count vehicles on the road, based on a prototype traffic management system. The paper also details the design and implementation of the prototype, using IoT and AI technologies to collect and process information about the flow of vehicles on public roads, using object detection with the YOLO algorithm. From this information, it communicates with other devices to coordinate the timing of traffic lights at each intersection, thus optimizing time. The trained model achieved a precision of 95% during training and 96% during testing. This research has the potential to be implemented by local governments in Latin America, improving decision-making to reduce traffic congestion.