ABSTRACT
Increasing traffic due to the increase in population and number of vehicles in metro cities calls for a smart traffic management system. Vehicle queues at traffic lights leads to an increase in petrol consumption and pollution. This paper proposes SmaTra, a smart traffic management system using hardware as well as a software approach. The hardware approach uses an array of Infrared Radiation (IR) sensors placed at the roadside which could be tripped by moving vehicles to indicate the density of traffic on each side of the road. This input from the IR sensors is used to change the vehicle waiting timing at the traffic junction. An alternate approach for smart traffic light using neural networks computation technique running on a neural network hardware engine is also developed. The hardware does image classification to identify the density of traffic and modifies the vehicle waiting timing at the junction. Out of the two approaches, the software approach gives more accurate and precise results up to an accuracy of 97% of identifying the traffic density.
