IJIRST –International Journal for Innovative Research in Science & Technology| Volume 4 | Issue 11 | April 2018 ISSN (online): 2349-6010 All rights reserved by www.ijirst.org 146 Real Time Automatic Helmet Detection of Bike Riders Kavyashree Devadiga Pratik Khanapurkar UG Student UG Student Department of Computer Engineering Department of Computer Engineering KKWIEER. Nashik KKWIEER. Nashik Shreya Joshi Shubhankar Deshpande UG Student UG Student Department of Computer Engineering Department of Computer Engineering KKWIEER. Nashik KKWIEER. Nashik Yash Gujarathi UG Student Department of Computer Engineering KKWIEER. Nashik Abstract Nowadays two-wheeler is the most popular modes of transport. However, because of less protection there is a high risk involved. As a solution to this, it is highly desirable for bike-riders to use helmet. Observing the usefulness of helmet, Governments have made it punishable offense to ride a bike without helmet and have adopted manual strategies to catch the violators which has limitations of speed. Using video surveillance of the street, the proposed approach detects if the bike rider is wearing a helmet automatically without manual help. If a bike rider is detected not wearing a helmet, the number plate of the vehicle read and noted. A database will be generated with records to identify every offender accurately. The system implements pure machine learning in order to identify every type of helmet that it comes across with minimum computation cost. Keywords: Helmet detection system, Feature evaluation and selection, Vehicle classification, Machine Learning, optical character recognition _______________________________________________________________________________________________________ I. INTRODUCTION According to the statistics given by transport ministry about 28 two-wheeler riders die daily on Indian roads in 2016 because of not wearing helmets. In the year 2017, 31 out of 100 people died in road accidents which shows increased rate from 21.6 deaths per 100 accidents in 2005. Also, it is proved that one of every five bike riders who died on roads were not wearing helmet. A study has estimated that not wearing a helmet increases the chances of death in an accidents by 42%. The number of deaths might be even more than the data which could not be collected. The deadliest fact is that out of the total deaths the average age was between 18 to 45 years which is a very productive age group. There are existing methods which uses specialized sensors in the ergonomics of the motorbike to check the presence of helmet. But it is impossible to convince every user for installation of sensors on the already existing bikes. Also, the accuracy and integrity of these sensors is questionable. Apart from this, systems that use video processing have very high computational costs. The technologies that were used to build the system very expensive hence making it an economically non-viable choice. This paper tries to mitigate the aforementioned problems by proposing a potential solution using continuous real time video processing of the traffic video feed. The proposed solution will provide a completely free of cost system once installed. Also, the software was build using free and open source technologies hence it has an overall software cost equal to zero. Every vehicle on the street will be evaluated and a database of the offenders will be generated real time. Hence given that every rider not wearing helmet is prosecuted, there will be an increased awareness in the public. II. BACKGROUND AND RELATED WORK The existing work that solves the problem by image processing solutions use technologies like HOG, LBP, WT [1][2][4]. The system proposed by isolates the bikes from images and by approximation crops the most probable area where helmet might be present and then feeds it to the feature extraction and matching system. Chiverton[1] proposed the use of circular arc to identify helmet in a video feed, it has very low accuracy. On the other hand, given the number of vehicles on the speed at a given instant, the computation that required is very heavy and consumes lots of resources. This method will determine any circular object around the bike rider as helmet