Computer vision based traffic monitoring system for smart cities

Description

Abstract

We have rules to govern traffic to ensure a smooth movement of traffic. We need to make sure that traffic violations do not occur and also penalize the offenders. We are considering the aspect of over-speeding for our implementation and would like to do our bit to supervise such an action. We receive the traffic surveillance video input from a camera. We have implemented a system which measures the number of vehicles and also checks for speed limit violation. The system has an accuracy of 98.96% for vehicle count detection and an accuracy of 98.14% for speed violation detection. We use the concept of multiple reference lines for identifying vehicles. Using the total duration of frames the vehicle took to cross the reference lines we determine the speed and hence the possibility of a traffic violation. It also helps in providing an idea regarding the traffic in any area of interest.