AI-Based Driver Drowsiness Detection

Authors:

Siddi Chaitanya, Dr. S.V.Hemanth, Koti Jyothsna, Ravula Sanjana, Kulkarni Vaishnavi, N Vishnu Vardhan

Abstract - Driver fatigue and drowsiness have been considered as leading causes of road accidents accounting for a large number of injuries and deaths annually. To prevent this type of problem, in this paper is proposed an AI-Based Driver Drowsiness Detection System that keeps track of the driver's alertness in real time leveraging computer vision along with machine learning techniques. The proposed approach uses a webcam to capture live video, then applies MediaPipe Face Mesh with OpenCV to extract facial landmarks. Using parameters like Eye Aspect Ratio (EAR) and Mouth Aspect Ratio (MAR), it is able to detect eye closure, blinking rate and yawning which is related to tiredness. When drowsiness overrides the threshold, an audio-visual alert system is activated to alert the driver. The proposed method achieves a favorable accuracy and computational expense making it promising for real-time applications on embedded or edge devices. The experimental results successfully validate that the system can accurately detect drowsiness in different lighting and environmental conditions and thus has the potential to improve driver safety and prevent accidents.

Keywords - Drowsiness detection, computer vision, CNN, Mediapipe, driver monitoring system, drunk driving, OpenCV, Eye Aspect Ratio (EAR), Mouth Aspect Ratio (MAR), Driver Monitoring System