HELMET PROTECTION DETECTION SYSTEM USING YOLOV8 FOR REAL-TIME TRAFFIC SAFETY MONITORING

Abstract: Road traffic accidents involving two-wheelers result in significant fatalities, with head injuries being the primary cause. Traditional manual helmet compliance monitoring suffers from limited coverage, human error, and scalability issues. This paper presents an automated Helmet Protection Detection System using YOLOv8, a state-of-the-art real-time object detection model. The system processes multiple input sources including static images, video files, live webcam feeds, and uniquely, YouTube livestreams. The YOLOv8 model is trained to classify motorcyclists into two categories: "Helmet" (compliant) and "No- Helmet" (violator), with color-coded bounding boxes and confidence scores. Experimental results demonstrate high detection accuracy with an overall 92%, precision of 90%, and recall of 87%. The system achieves real-time performance of 45 FPS on a standard GPU, making it suitable for live traffic monitoring. This cost-effective solution leverages existing camera infrastructure, reduces dependency on manual supervision, and contributes to enhanced traffic safety enforcement.

Keywords: YOLOv8, Helmet Detection, Real-Time Monitoring, Deep Learning, Traffic Safety, Computer Vision.