International Journal of Scientific Research and Engineering Development

Driver drowsiness is a leading cause of road fatalities worldwide. This paper presents SafeDrive Monitor, a real-time driver drowsiness detection system developed using Python, Flask, OpenCV, and Google MediaPipe Face Landmarker. The system continuously captures webcam frames, extracts 478 3D facial landmarks, and computes the Eye Aspect Ratio (EAR) bilaterally to determine eye closure state. When the EAR remains below a threshold of 0.25 for 2.5 consecutive seconds, the system classifies the driver as drowsy and triggers an audible alarm. A Flask-based web interface streams the annotated live feed and provides session control via start/stop routes. Evaluation on a CPU-only machine demonstrates approximately 28 frames per second with alarm latency under 2.7 seconds. The system requires no specialized hardware, offering a cost-effective, deployable road safety solution. Limitations including platform-specific audio and lighting sensitivity are identified, and directions for future work are discussed.

Sanjay R, Dr.K.Thenmozhi,"A Real-Time Driver Drowsiness Detection Systems" International Journal of Scientific Research and Engineering Development, V9(2): Page(548-554) Mar-Apr 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.