DROWSINESS DETECTION SYSTEM USING MACHINE LEARNING

Driver drowsiness remains a critical contributor to road accidents, largely due to reduced alertness, slower reaction time, and impaired decision-making during prolonged driving. Modern driver-assistance systems often rely on indirect performance-based indicators, yet they fail to capture clear physiological symptoms of fatigue. To address this limitation, this study presents an intelligent, camera-based driver monitoring system that analyzes visual cues such as eye closure duration, blink rate, head rotation, and yawning frequency. A large dataset collected through controlled driving simulations is used to extract multiple facial and ocular features using advanced image-processing techniques and shape-predictor models. These features are then evaluated through machine learning–based classification methods to distinguish between awake, mildly fatigued, and drowsy states. A fusion approach combining eye and mouth behavior significantly enhances detection reliability. The proposed system is tested in both offline and real-time environments, demonstrating strong accuracy and robustness in identifying early signs of fatigue. By providing timely warnings to the driver and enabling remote monitoring through IoT integration, the system contributes toward reducing fatigue-induced accidents and improving overall traffic safety.