Smart Helmet Detection System for Road Safety Using YOLOv8

Helmet compliance has emerged as a major public safety issue as a result of the growing number of motorcycle accidents triggered by helmetless riders. In this paper, a Smart Helmet Detection System employing YOLOv8 is proposed and tested with the aim of promoting motorcycle safety through computerized, real-time surveillance. Road accidents result in the death of nearly 1.35 million people annually, and motorcyclists are particularly vulnerable. The proposed strategy evaluates helmet use that is appropriate for the circumstances, such as whether it is being worn correctly, under various environmental conditions, unlike traditional systems that only identify the presence of a helmet. Regarding real-time detection and reliability, YOLOv8 is better than earlier models with an accuracy of $75 \%$, recall of $78 \%$, and system accuracy of about $70 \%$ using a labelled set. Detection of non-helmeted individuals is still challenging, however, which means that classic approaches may still work a little better under certain regulated conditions. It is hoped that additional enhancements-such as increasing the dataset, tuning hyperparameters, and merging with the real world-will enhance detection accuracy and adaptability even further. This system demonstrates how AI-powered surveillance can enhance traffic law enforcement, reduce the number of traffic accidents, and promote helmet usage.