DefinePK

Abstract

Helmets are essential safety measures for anyone using two-wheeled vehicles (motorcycles, bicycles, and e-scooters), and the absence of protective helmets may result in serious or fatal injuries. The principal technique for helmet detection is now a series of Convolutional Neural Network methods. Detection accuracy, speed prediction, and ease of deployment are essential criteria for achieving road safety. Conventional object identification methods often fail to provide consistent performance across all domains. This study presents a helmet identification application using the latest You only look once version 7 (YOLOv7) algorithm enhanced by an attention-based approach. The model's performance was assessed using a collection of helmet test photos, achieving an average accuracy (mAP@0.5) of 91.4%. The findings demonstrate great detection accuracy and minimal computing requirements, making the model appropriate for practical use. Consequently, the suggested model may aid in addressing the issue of helmet detection on two-wheeled vehicles.