Vehicle Data Aggregation from Highway Video of Madurai City Using Convolution Neural Network

Abstract

Abstract Traffic is one of the key issues in highly populated countries like India. There are many research works going on in the areas of the travel and transportation sector. These sectors having challenges such as traffic monitoring, routing assistance, personalized driving. Nowadays, smart city is one of the Government mission which includes traffic management, crowd management etc. In the recent past, there are many cities received funds from the government of India under the smart city scheme. In this paper, the vehicle data aggregation process is proposed using deep Convolution Neural Network (CNN) particularly, in and around the areas of Madurai city, Tamilnadu, India. The purpose of the location chosen is Madurai city also received funds under a smart city scheme from Indian government. There are certain works are already going on in Madurai city under the smart city project. The real-time traffic video capturing used to raspberry pi B modules camera. To support further into the work, a data aggregation technique is proposed in this work to create a vehicle dataset for the Madurai city to help traffic management, driver assistance, and routing guidance through automation system. In this work, the CNN algorithm detected the different vehicles with 92.9% accuracy. This will be compared with region based CNN algorithm, which detected with 96.4% accurately. Hence it is proven that the region based CNN performs better in vehicle detection. The vehicle data are tabulated from some of the sample videos captured in and around Madurai highways using CNN through MATLAB.