The Design and Implementation of a Drowsiness Detection and Alarm System

Drowsy driving poses a significant and increasing threat to road safety, leading to a high proportion of traffic accidents and substantial economic burdens. This paper presents the design and implementation of an innovative drowsiness detection and alarm system aimed at mitigating these risks. The proposed system employs a camera module for real-time facial monitoring and leverages an AI model, specifically Google's MediaPipe FaceLandmark, to analyze key indicators such as eye blink frequency, duration of eye closure, and head movements [2]. By integrating image preprocessing techniques like histogram equalization and a sophisticated drowsiness detection algorithm based on Eye Aspect Ratio (EAR) and head pose estimation, the system accurately identifies various states of driver drowsiness [1], [3]. Upon detection, it activates an audible alarm via a buzzer controlled by Raspberry Pi's GPIO pins, effectively alerting the driver. Experimental results demonstrate the system's efficacy in detecting all predefined drowsy states—frequent blinking, head nodding, head drooping, and prolonged eye closure—and triggering the alarm successfully within a vehicle environment. This technology offers diverse applications beyond driving, including industrial safety, academic efficiency, healthcare monitoring, and home safety, promising significant improvements in accident prevention, productivity, and overall quality of life [4], [5], [6], [7]. Future work will focus on algorithm refinement, testing in varied environments, and incorporating user feedback to enhance accuracy and user comfort [8], [9], [10].