Traffic Light Negotiation Using Image Processing

In urban environments, increasing traffic congestion demands smarter and more responsive control systems. This project proposes a traffic signal management system that dynamically adjusts signal timings based on real-time vehicle density using advanced image processing and object detection. We employed the YOLOv8n model, known for its speed and accuracy, to detect and count vehicles from live video feeds captured via IP webcams. By analyzing lane-wise traffic density through edge detection and bounding box regression, the system intelligently assigns green light durations to the busiest lanes, significantly reducing overall wait times and improving traffic flow. Our approach offers faster response compared to traditional fixed-timer systems and demonstrates high accuracy even under variable traffic conditions. Experimental simulations using Pygame visualize signal switching based on real-time vehicle counts, validating the system's practical effectiveness in smart city scenarios.