Automatic Detection of Bikers with No Helmet and Number Plate Detection

Abstract

In a very highly populated country like India enforcing traffic rules becomes a difficult task. Many people ride motorcycles without helmets and doing so, they are risking their life. Monitoring of all vehicles by traffic police is a very tedious task and there is a need for automation for such tasks. Many of the heavily populated cities are employed by CCTV's. The available infrastructure can be used to automate the monitoring of motorcyclists. In this paper a method to detect motorcyclists without Helmet and extracting the number plate of the motorcycles whose rider is not wearing helmet is proposed. The method uses the YOLO (You Look Only Once) model to detect the helmet and license plate of the motorcyclist and motorcycle respectively. YOLO is a model which is trained on COCO dataset and is capable of detecting 80 classes. The YOLO model was trained on Helmet and number plate datasets. The results achieved by this method are very good. The accuracy is high in the detection of helmets and license plates.