This study introduces an AI-driven Traffic Flow Management System designed to enhance urban mobility and safety by addressing critical issues like congestion, vehicle-induced blockages, and road infrastructure defects. The solution integrates real-time vehicle counting, blockage detection based on the traffic flow rate equation, emergency vehicle prioritization, and crowdsourced pothole detection. A core feature is the system’s ability to dynamically adjust the duration of IoT-controlled traffic signals in response to detected blockages, optimizing traffic throughput. The system employs computer vision models for detection tasks. The Emergency Vehicle Detection model achieves 94.5% precision, 92.9% recall, and 96.1% mAP@50. Pothole detection attains 80.8% precision, 87.2% recall, and 88.5% mAP@50. The dynamic signal control algorithm proved effective in managing simulated traffic anomalies. The architecture is scalable, supporting real-time processing and sub-5-minute emergency response capabilities. The findings demonstrate the viability of a data-centric, integrated system for achieving proactive traffic control, offering a scalable blueprint for managing dynamic urban traffic and improving both safety and efficiency.