Home / Articles
| DRIVER DROWSINESS DECTECTION SYSTEM |
|
|
Author Name Ms. J. Aiswarya and Sankari k Abstract The Driver drowsiness is a significant contributor to road traffic accidents worldwide, frequently leading to serious injuries, fatalities, and substantial economic losses. Fatigue resulting from prolonged driving hours, insufficient sleep, and psychological stress impairs cognitive performance by reducing alertness, slowing reaction time, and weakening decision-making capability. As drowsiness often develops gradually and unnoticed, early detection is crucial for preventing accidents and enhancing overall transportation safety. This paper presents a Real-Time Driver Drowsiness Detection System that leverages Convolutional Neural Networks (CNN) and advanced computer vision techniques to automatically identify fatigue- related behavioral patterns. The proposed system continuously captures live video streams through a camera and performs facial landmark detection, eye region extraction, and temporal eye-state analysis. Key parameters such as eye closure duration (PERCLOS), blink frequency, and yawning detection are evaluated to assess the driver’s alertness level.
The CNN model is trained on publicly available eye-state datasets combined with real-time captured facial images to improve classification robustness under varying lighting and facial conditions. The system processes video frames in real time, applies preprocessing techniques including grayscale normalization and noise reduction, and classifies the driver’s state as either alert or drowsy. Upon detecting prolonged eye closure or abnormal blink patterns, the system immediately triggers audio and visual warning signals to regain driver attention. Experimental analysis demonstrates high classification accuracy, low inference latency, and stable performance in real-time conditions. The solution is non-intrusive, cost-effective, and scalable, making it suitable for integration into smart vehicles and Advanced Driver Assistance Systems (ADAS). By combining deep learning and behavioral analysis, the proposed framework offers an efficient and practical approach to reducing fatigue-related road accidents and improving transportation safety standards. Keywords: Keywords: Driver Drowsiness Detection, Convolutional Neural Network (CNN), Deep Learning, Computer Vision, Eye State Classification, PERCLOS, Facial Landmark Detection, Real-Time Monitoring, Machine Learning, Road Safety, Advanced Driver Assistance Systems (ADAS). Published On : 2026-03-07 Article Download : 
|
|
