Drowsiness Detection Using Convolutional Neural Network

Abstract— Driver drowsiness is one of the major causes of road accidents, especially during long-distance and night-time driving. Continuous monitoring of driver alertness using non- intrusive techniques has therefore become essential for improving road safety. This paper presents a real-time drowsiness detection system based on facial feature analysis using a Convolutional Neural Network (CNN). The proposed system captures live video through a standard webcam and processes facial features such as eye closure and yawning to determine the driver’s alertness level. Facial landmarks are extracted to compute Eye Aspect Ratio (EAR) and Mouth Aspect Ratio (MAR), which serve as key indicators of fatigue. A multi- scale CNN architecture is employed to learn both fine and coarse facial patterns, improving detection accuracy under varying lighting conditions and face orientations. When drowsiness is detected, the system immediately generates visual and audio alerts to warn the driver. The experimental results demonstrate that the proposed approach operates effectively in real time using low-cost hardware, making it suitable for practical deployment in intelligent transportation and safety systems. Keywords— Drowsiness Detection, Convolutional Neural Network (CNN), Facial Feature Analysis, Eye Aspect Ratio (EAR), Mouth Aspect Ratio (MAR), Real-Time Monitoring, Driver Safety.