WAKE-UP YOLO: A Real-Time Multi-Modal Driver Drowsiness Detection System Using YOLOv8n and Physiological Sensor Fusion

Road accidents caused by driver fatigue and drowsiness remain one of the leading causes of fatalities worldwide. Existing warning systems often rely on single-source data, which results in false positives and missed detections. This project, WAKE-UP-YOLO, presents a comprehensive real-time driver drowsiness detection system that integrates computer vision using the YOLOv8n deep learning model with physiological health monitoring to provide reliable, multi-modal drowsiness alerts. The system performs real-time detection of facial indicators such as prolonged eye closure and yawning through a camera feed processed by YOLOv8n. Simultaneously, it monitors health parameters including pulse rate, blood oxygen saturation (SpO2), and stress level sourced from a smartwatch API or a simulation engine. A weighted fusion logic module combines both signal streams into a unified drowsiness score. When the score crosses configurable thresholds, the system triggers audio alerts at warning (score >= 0.6) and critical (score >= 0.8) severity levels. The system is implemented in Python using OpenCV, Ultralytics YOLOv8, Pygame, and Tkinter, and is deployable on standard laptops with webcam access. Experimental results demonstrate that multi-modal fusion significantly reduces false positives compared to single-source systems, making WAKE-UP-YOLO a practical and effective driver safety solution.