Driver Drowsiness Detection Using Multi- Channel Second Order Blind Identifications

Abstract: This project presents a comprehensive, real-time driver drowsiness detection and alert system using facial landmark analysis, remote photoplethysmography (rPPG), and machine learning. The system captures live video through a webcam and extracts key features such as Eye Aspect Ratio (EAR), Mouth Aspect Ratio (MAR), head pose angles, and heart rate using non-contact methods. A deep learning model processes these indicators to accurately classify driver alertness. Upon detecting drowsiness, the system immediately triggers audio-visual alerts to regain driver attention. Designed for non-intrusive monitoring and ease of deployment, this tool aims to enhance road safety and reduce fatigue-related accidents. The system also includes a GUI for usability and can be extended with additional safety interventions. Index Terms-Driver Drowsiness Detection, Eye Aspect Ratio (EAR), Mouth Aspect Ratio (MAR), Head Pose Estimation, Heart Rate Monitoring, Remote Photoplethysmography (rPPG), Deep Learning, Support Vector Machine (SVM), Real-Time Alert System, Facial Landmark Detection, Driver Safety.

Keywords: Detecting Driver Drowsiness