Advanced Computer Vision-Based Surveillance System for Helmet Detection and Triple Rider Identification on Motorcycles using Machine Learning

Abstract

In many countries, the number of motorbike accidents has been steadily rising over time. Over 37 million individuals in India own and operate two-wheelers. As a result, it's imperative to create a system for automatically detecting triple riding and helmet use. So, a machine learning-based technique is used to construct a unique object detection model that can recognize motorcycle riders. When a helmetless rider and triple rides are detected, the number plate is extracted, and the number plate number is recognized using an optical character recognition device. If the bike riders weren't wearing helmets and if they were performing triple rides, then this was the case. With the use of a webcam, this application can be used in real time. With the help of deep learning, we can categorize the image of the three cyclists in this project. This uses a CNN-based architecture to categorize the image. The initial step in this project's approach is to train the webcam's captured image. The network will then provide the output based on the trained data by taking into account webcam input. 70% accuracy is what this will yield.