Published Paper Details:

This research presents a real-time driver drowsiness detection and alert system using Python-based computer vision and machine learning techniques. The system combines several detection methods--including Eye Aspect Ratio (EAR), facial landmark tracking, head pose estimation, and temporal behavioral analysis--to accurately identify signs of fatigue. Using OpenCV, dlib's 68-point landmark model, and scipy for signal analysis, it provides multi-modal alerts through audio, visual, and haptic feedback. Tested on 150 participants under varied lighting, demographics, and driving conditions, the system achieved 96.7% accuracy, 94.3% sensitivity, 97.8% specificity, and a rapid 0.98-second response time. It also maintains real-time performance at 35 FPS and reduces false positives to 2.1%, outperforming existing commercial solutions. Additional features include adaptive user calibration, environmental condition adjustments, and APIs for seamless integration into vehicle systems. Overall, the system demonstrates strong reliability and suitability for widespread use in advanced driver assistance systems (ADAS).