Intelligent traffic management:

Abstract

The efficient management of modern traffic systems increasingly relies on automated technologies for real-time vehicle detection. This research article studies the development of an intelligent traffic management system, featuring an emphasis on real-time vehicle identification and traffic sign allocation using the YOLOv7 (You Only Look Once) object detection framework. The research examines the crucial need for effective traffic management solutions in metropolitan environments, focusing on the crowded roadways of Dhaka, Bangladesh. A large dataset of 1760 photos of Dhaka's busiest streets was used to train and assess the YOLOv7 model. The trained model scored an astounding 93.77% accuracy, indicating its usefulness in real-time vehicle recognition. Furthermore, the YOLOv7x variation, which has a more sophisticated design, was tested, resulting in an accuracy of 81.6%. The research emphasizes the significance of accurate and rapid vehicle detection in enhancing traffic flow and safety through the strategic allocation of traffic signs. By identifying and analyzing vehicle patterns and traffic density in real-time, the YOLOv7 model can facilitate dynamic adjustments to traffic signal timings and the strategic placement of traffic signs, effectively reducing congestion and improving overall traffic efficiency. This capability is crucial for urban centers like Dhaka, where traffic congestion is a persistent issue. This study provides a detailed analysis of the training process, model performance, and the practical implications of deploying such an intelligent system in a metropolitan context. The comprehensive evaluation includes a discussion on the strengths and limitations of the YOLOv7 and YOLOv7x models, as well as their suitability for real-time traffic management applications. The findings underscore the viability of YOLOv7 as a robust tool for intelligent traffic management, paving the way for further advancements in smart city infrastructure and automated traffic control systems. Additionally, this research highlights the potential for future enhancements and integration with other smart city technologies, contributing to a more efficient and sustainable urban traffic ecosystem.