Deep Learning Based Detection and Passive Enforcement System for Helmet in Two Wheelers

Abstract: In this paper an attempt is made to reduce causality in the two wheeler accidents, which contributes more than 40% of the total number of road accidents. It is observed that awareness and execution of the traffic rules plays a vital role. Since the number of accidents is increasing every year, any attempt using the latest technological development can save hundreds of lives. Even though it is well known fact that the use of helmet can reduce the head injuries in the accidents, the negligence to use the same invites serious causalities in the two wheeler accidents. The proposed model presented in this paper use the potential of Artificial Intelligence (AI) and an effective implementation of the law to wear helmets for two wheeler passengers is ensured. Presently the law is executed with the strict actions of the traffic enforcement system comprising Police Department and Motor Vehicle Departments of the State. This physical system, address lot of limitations and two wheeler riders often violate the law in the rural roads. AI based smart system, captures sufficient number of images of the passengers wearing different helmets available in the market. These data is used to train the smart system and validated for standard conditions. The outcome of the AI is similar to the decision taken by the human brain, where neurons are acting as the nods. The built in camera placed in the front part of the vehicle capture the images in a sampling time interval of 800 ms. The images are processed to identify the status of the helmet and the response is used to control the speed of the vehicle. The absence of the helmet forces the maximum speed of the vehicle as 20km /hr so that the passenger will voluntarily use the helmet. The experiments shows 86% reliability and it is expected to be improved by the use of high resolution camera and modified algorithms used in the AI system.