Helmet detection using Image Processing and Deep Learning in Workplace

Abstract

Ensuring compliance with mandatory helmet laws is critical for workplace and road safety. Traditional enforcement methods, which rely on manual inspections and surveillance personnel, are inefficient, error-prone, and challenging to scale with increasing vehicle and rider volumes. Additionally, factors such as poor image quality, varying viewing angles, and inconsistent monitoring further hinder effective enforcement. To address these challenges, this study proposes an automated helmet detection and motorcycle license plate recognition system leveraging deep learning techniques. A Convolutional Neural Network (CNN) is employed to classify riders as helmeted or non-helmeted using grayscale pixel data from preprocessed images. Optical Character Recognition (OCR) is integrated for automatic extraction of motorcycle license plate numbers, enabling real-time identification of violators and YOLO a state of the art real time object detection algorithm that can quickly identify objects like riders and helmets in images. The system processes images from surveillance cameras, ensuring faster, more accurate, and consistent enforcement while minimizing human intervention. By automating helmet law compliance monitoring, the proposed solution enhances enforcement efficiency, accelerates penalty processing, and improves overall safety by reducing injuries and fatalities caused by non-compliance.