Road accidents are one of the major causes of human deaths. Among the different types of road accidents, motorcycle accidents are common and cause severe injuries. The helmet is the motorcyclist's main protection. Most countries require the use of helmets by motorcyclists, but many people fail to obey the law for various reasons. We present the development of a system using image processing and deep convolutional neural networks (CNNs) for finding motorcyclists who are violating helmet laws. The system comprises motorcycle detection, helmet vs. no-helmet classification, and motorcycle license plate recognition. We evaluate the system in terms of accuracy and speed. The system has been deployed at several locations in Bangkok and Phuket, Thailand since 2016. Early reports indicate that compliance with motorcycle helmet laws is increasing.
Helmet violation processing using deep learning
2018-01-01
160750 byte
Conference paper
Electronic Resource
English
Deep-learning based helmet violation detection system
IEEE | 2021
|Traffic Rules Violation Detection using Deep Learning
IEEE | 2020
|Detecting motorcycle helmet use with deep learning
Elsevier | 2019
|Deep learning based vehicle violation detection system
IEEE | 2021
|