Urban regions are growing at a very high rate and this causes traffic congestion, which is a persistent challenge for people and the environment. This leads to longer travel times, higher fuel consumption, and environmental degradation. It also has the potential to impede emergency response systems, risking lives in the event of emergencies. Conventional traffic optimization techniques struggle with real-time adaptability and the existing available AI technologies have some limitations. Limitations include multi-intersection coordination challenges and lack of integration for congestion caused by accidents and large pedestrian flows. To address these limitations, this research introduces the Hybrid Quantum-Inspired Multi-Agent Reinforcement Learning (HQMARL) model. This is an advanced framework that integrates the Multi-Agent Reinforcement Learning (MARL) for local traffic control with the Quantum Approximate Optimization Algorithm (QAOA) for global coordination. The system uses computer vision for vehicle detection. This helps in optimizing the traffic signals according to the real-time traffic density. It prioritizes emergency vehicles and allows adaptive adjustment of signals to mitigate congestion if an accident is detected. A Crowd Mode is also introduced to optimize signal durations during large-scale public events to facilitate free traffic flow. SUMO simulations demonstrate that HQMARL minimizes congestion. It also enhances the efficiency of the emergency response system. Therefore, HQMARL produces a scalable and adaptive solution for intelligent traffic control by integrating computer vision, quantum-inspired optimization, and reinforcement learning.