Developing a Machine Learning System to Automatically Control Traffic Signals

Abstract

With the advancement in technology, we have seen many things evolve and get better with the help of technology. Traffic signals is not one of those things. In this proposal we suggest a method that would use image processing to send information about traffic congestion to the AI model that would control the signals based on that. Our objective is to make traffic signals smart. With the use of technology, it is possible to detect and take decisions by the traffic signal, just like a traffic warden does, and it would remove the human error. It would prioritize emergency vehicles eliminating risk of accidents using real-time decision making. This would really improve experience for people waiting at traffic signals and in most cases help save time for people. This project aims to design and control a Traffic Automatically for Traffic signals. In this first phase, we will use YOLOv5 to work on live videos. To build an automatic traffic system, a huge amount of data is needed. In the second phase, we will implement our System. In the final phase, we will train and test our Model. This model will be based on YOLOv5 and an algorithm to give best outcome. The evaluation will be carried out using Accuracy.