Neural Network Based Smart City Application for Traffic Violation Detection

Abstract

The main application of helmet detection is in traffic roads where accidents are more. Even though various measures are taken by government, it is not followed correctly by the motorcyclists, so several smart techniques should be employed. In developing countries like India, the two-wheeler is the most common means of transportation. Though it is convenient to ride, negligence of Helmet compulsion law by the riders is leading to many accidents. According to the statistics provided in the Road Accident Report, at least 98 riders, who were not wearing a helmet, died daily in 2017. Deaths caused by not wearing helmet rose to 36,000 in 2017 which was 10,135 in 2016. WHO has declared that the negligence of riders in using safety devices as one of the causes for the rise in road accidents. Government has implemented many methods to catch the violators. But those methods require human assistance, which decreases the performance and reliability of the system. In this project, an automatic helmet detection system is designed using Faster RCNN machine learning algorithm. Faster RCNN algorithm is used to detect the helmet.