AI Enabled Next-Generation Traffic Control System

Abstract

Traffic is one of the superior problems in modern metropolis. Fresh and advanced technology related infusions are required to supervise themselves and direct traffic signals in order to decrease the snarl-upping of traffic. Major problem is when it comes to a predicament or an emergency circumstance which affects the servicing facilities like ambulances, fire trucks, police vans etc. In this paper, we capture data from the surveillance camera and using it we will train the machine using Machine Learning and Deep Learning. So, the process goes where we use a collective number of images which can be enormous in numbers which can be used to train the model. Subsequently, the vehicles are identified, and are categorized into various classes and this classification is done by itself, as it is edified to precision. We procured 88% accuracy using YOLOv5 for vehicle recognition. Further it contributes to the future, so that road design and scrutiny can be developed and secondly the fuel usage can be controlled, and the standby time is also saved effectively. Within some period, we will be able to harmonize most of the signals, by imparting a flexible traffic management system, thus resulting in declination of traffic congestion.