Over the years, motorcycle accidents have increased in various countries. Motorcycles are growing more popular as a result of many social and economic causes. Although the use of helmets is made mandatory in many countries for motorcyclists, most of them do not wear a helmet. A motorcycle accident might be fatal if the rider is not wearing a helmet. Detecting such offenders of traffic rules is a highly desirable but necessary task to ensure safety measures due to many obstacles such as occlusion, illumination, poor quality surveillance video, fluctuating weather conditions, and so on. This paper aims to explain and illustrate a framework for identifying license plates of motorcyclists who ride them without helmets in surveillance videos. In the proposed approach, we generated a dataset from a real-time surveillance video that is turn fed to our custom deep learning model using the YOLOV3 framework, which comes under the class of single-shot detector algorithm for object detection. We detected multiple objects for every image and recognized license plates based on whether the motorcycle rider wore a helmet or not. Object detectors are evaluated using mean average precision, and our evaluation results are 68.79% with an IoU value of 70%.
CITATION STYLE
Devi, S. K. C., Reddy, G. V., Aakarsh, Y., & Gowtham, B. (2023). License Plate Detection of Motorcyclists Without Helmets. In Lecture Notes on Data Engineering and Communications Technologies (Vol. 142, pp. 295–301). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-19-3391-2_22
Mendeley helps you to discover research relevant for your work.