A Review on Deep Learning Techniques for Detecting Driver Fatigue

A Review on Deep Learning Techniques for Detecting Driver Fatigue

Prof. Shital Aher1, Bhakti Dattatray Shinde2, Sarita Rambhau Sadgir3, Rutuja Anil More4, Rutuja Kishor Jadhav5
1Professor, Department of Information Technology, Sir Visvesvaraya Institute of Technology, Maharashtra, India

2,3,4,5Department of Information Technology, Sir Visvesvaraya Institute of Technology, Maharashtra, India

Abstract - This project proposes an intelligent driver monitoring system designed to enhance road safety by detecting drowsiness and alcohol impairment. It uses visual information to track facial features, focusing on eye and mouth movements, and calculates the Percentage of Eye Closure (PERCLOS) — a proven metric for detecting fatigue. Haar cascade algorithms are employed for real-time detection of eyes and mouth, enabling early identification of fatigue signs like prolonged eye closure and yawning. Deep learning techniques, particularly the Softmax function, are used for accurate classification of drowsiness levels. Additionally, the system integrates alcohol pulse detection to determine driver impairment. By issuing timely alerts, the system aims to reduce accidents caused by fatigue and intoxication, especially among drivers of long-haul and heavy vehicles.

.

Key Words: Driver Drowsiness Detection, PERCLOS, Haar cascade, Deep Learning, Softmax Function, Alcohol Detection, Facial Recognition, Road Safety, Fatigue Monitoring, Real-Time Alert System