Traffic Density Detection And Signal Optimization Using YOLOv5 And ANT Colony Optimization

Urban traffic jam leads to delays, wastage of fuel and air pollution, and fixed-time traffic signals do not react to real-time scenarios. The findings of this study recommend detection of cars using YOLOv5, regression methods for finding the optimum duration of green light, and reinforcement learning (DQN) for making adaptive signal control decisions. Traffic flow prediction is improved by LSTM models. Performance is quantified in the form of vehicle clearance rate, waiting time reduction, and traffic flow rate. Experiments show that the adaptive system outperforms standard fixed-timing methods greatly, with YOLOv5 detection being 60% accurate, DQN lowering waiting times by 37.2%, vehicle clearance rates increasing by 42.8% during rush hours, and carbon emissions lowering by 27.4%, hence a scalable model for smart cities.