Author Email: email@example.com
Nowadays, motorcycle accidents is on high rise and the main cause is traumatic brain injuries. Which suffers because most of the motorcyclists do not use helmet. In this paper we proposed a frame work for automatic detection of motorcyclists without helmet on public roads using surveillance videos. The frame work uses (SVM) support vector machine trained on features derived from the upper 60% part of the motorcyclists using Daubechies 8 wavelet transform. Then upper 60% part of the motorcyclists is isolated and then classified as helmet versus non – helmet using trained classifier. The performance of the proposed approach was evaluated on static photographs and videos. Experiment on real videos successfully detect 95.23% of traffic rule violators and 97.33 %on static photographs.
Helmet Detection, Daubechies 8 Wavelet Transform, Saliency Map, (SVM) Support Vector Machine.
In India, over 78% of vehicles on the road are two-wheelers and they account for about 31.5 % of road accident deaths, a statistic that has risen steeply over the years from 24.3% in 2014 to 26.8% in 2015 and 28.8% in 2016.Except two wheelers other road accidents are those in cars, taxis, van and other light medium motor vehicle (17.2%), trucks (11.4%), pedestrians (9.3%) etc . Dr. Mc Misra , director and dean of AIIMS says “Wearing a helmet and tying it properly can prevent loss of lives by 90 percent in motorcycle accident cases. We call it helmet vaccine,” . Helmet is safety armor for head, just like skull protects the brain from impacts helmet protects the head. The three magical layers of helmet thermoplastic layer, expanded polystyrene (eps) and inner soft spongy layer protect user from traumatic brain injuries . National highway of traffic safety administration report says “The ratio of two-wheelers on public roads is 50% to that of human population”. According to section 129 of the motorcycle vehicle act Government has made it mandatory to ride a motorcycle with helmet . Nowadays traffic cops are capturing the photographs of the people not wearing helmet and necessary actions are taken against the violators but in this scenario human preference get the first priority .So automation of this process is on high requirement. And in the current scenario there are CCTV surveillance in every country so, implementation will also be cost effective.
In this paper we proposed an approach for automatic detection of bike riders with and without helmet which will help the traffic police for detecting such violators and take necessary action against them. In the proposed has met certain challenges related to poor quality of the videos and environmental conditions. One future work can be the automatic detection of these violators and if somebody is found violating the rules then trace the license number plate of the vehicle and automatic actions to be taken. No manual work should be done which will in turn increase the efficiency and time awareness among the violators.
223 total views, 1 views today