Abstract:
Traffic congestion remains a pervasive issue in metropolitan contexts. Demanding effective solutions for traffic management and road safety enhancement. Despite breakthroughs in technology, Current technologies for vehicle speed detection and traffic control generally suffer from limits in accuracy and efficiency. To address this gap, our work intends to design a robust and efficient system for real-time traffic control utilizing machine learning approaches. Leveraging image processing and artificial neural networks (ANNs). We put forward a technique which can accurately distinguish among the different speed vehicles in a traffic. The approach has three main components: pre-processing, identifying features, and classification by ANNs. We relied on a multimedia dataset which was retrieved from a stationary camera that recorded scenes from the roadway to train and test our algorithm. The result validation indicated the high efficiency of a method by a way of the algorithm performance reaching accuracy of 92.5%, precision of 89.3%, recall of 94.7%, F1-score of 91.8%, and MCC of 0.86 on the validation set. In addition to the simulation-based tests, it was also reported that the models displayed stability and efficiency in real-world traffic control environments, thus, proving that they can be used to manage traffic, hence improving road safety. Through this research study, not only we are trying to combat the already present issues related to traffic management but also by providing the basic framework, we make it possible for further improvements in modern transportation systems.