This article proposes an Intelligent Traffic Light Management System (ITLMS) to manage traffic flow in urban environments, improve the green wave, and prioritize emergency vehicles, through a combination of edge computing and advanced deep learning technologies. ITLMS incorporates You Only Look Once version 8 medium (YOLOv8m) to accurately and in real time classify multiple vehicle classes, including ambulance, firefighter, police, car ( i.e ., normal vehicle), and traffic jam, enabling emergency responders to pass through intersections as quickly as possible with minimum delays. The proposed ITLMS is based on an Internet of Things (IoT) architecture, where edge nodes are deployed at intersections to process the data locally in real-time and make immediate adjustments to traffic lights. The cloud is also used to analyze overall traffic patterns and coordinate traffic light changes across multiple intersections, forming a smart network that can react in real-time to traffic conditions. We also built our own dataset of 5,500 images containing emergency vehicles ( i.e ., ambulance, firefighter, and police) from 11 countries, multiple domains, and languages to illustrate and evaluate ITLMS. Experiment results show that the YOLOv8m model outperforms the others (YOLOv9c, YOLOv10m, and Real-Time Detection Transformer (RT-DETR)) in terms of accuracy and computational cost for this task. Specifically, YOLOv8m achieves a mean Average Precision (mAP50) of 0.947 and an F1-score of 0.919. Its accuracy, precision, and recall are 0.947, 0.916, and 0.923, respectively. Although the accuracy of YOLOv8m on the edge node is 8.8 ms, which is higher than YOLOv9c’s 6.7 ms, YOLOv8m is more precise. It offers the best trade-off between accuracy, precision, and recall, making it best suited for real-time traffic management in urban settings. ITLMS operates in milliseconds, can adapt to traffic congestion, and prioritizes emergency vehicles without requiring additional sensors, making traffic safer and more efficient in cities. This study presents ITLMS as a solution that is not only scalable and flexible but can also be easily integrated into existing traffic infrastructure to enable efficient emergency response and traffic flow in urban areas. Under real deployment settings, detection performance can be impaired by issues such as camera placement, occlusion, and bad weather conditions. To mitigate these issues, ITLMS employs data augmentation, adaptive calibration, and fallback control to enhance robustness.
Noor et al. (Thu,) studied this question.