Real-Time Adaptive Traffic Control System with Emergency Vehicle Detection Based on Computer Vision and Sound Frequency Monitoring

Efficient traffic management in smart cities is crucial, especially for emergency vehicles. This paper presents a Real-Time Adaptive Traffic Control System that integrates YOLOv3 object detection algorithm to analyze real-time video streams from cameras, identifying vehicles and calculating traffic density based on vehicle count and lane occupancy achieved through NodeMCU1 microcontroller. Microphone INMP441 is used to transmit digital audio data via the I2S protocol to a NodeMCU2 microcontroller. The captured sound is processed using the Fast Fourier Transform (FFT) algorithm to isolate siren frequencies within 500–2000 Hz range, as per Indian government regulations. By combining visual detection and sound frequency monitoring, the system utilizes a dual-layer verification method which reduces false positives and optimize the traffic flow leading to be suitable for smart cities application. Experimental results show that emergency vehicle detection accuracy reached 95% under ideal conditions. At intersections, the system led to a 30% reduction in average waiting times and improved emergency vehicle response times by 40% as compared to conventional systems. However, I2S data errors significantly impacted system performance, reducing accuracy by 50%. Hence, the proposed work is a potential method to be deployed in smart cities for traffic control and emergency vehicle detection.