A Cost-Effective Driver Drowsiness Detection Using Embedded Vision Systems Using Raspberry Pi for Enhanced Road Safety

Driver drowsiness is a critical contributor to road accidents globally, with significant fatalities reported in India due to sleep-deprived driving. This paper presents a costeffective, real-time drowsiness detection system leveraging computer vision and hardware integration to enhance road safety. The proposed system employs the Eye Aspect Ratio (EAR), computed using facial landmarks detected via the Dlib and OpenCV libraries. EAR thresholds categorize driver states into active, drowsy, and sleepy, triggering visual alerts (LEDs) and auditory alarms (buzzer) for timely intervention. Experimental results demonstrated $\mathbf{9 6 \%}$ accuracy under normal conditions, though performance dipped to 76 % in dim lighting or for person wearing glasses. Comparative analysis highlighted the efficacy of real-time processing, portability, cost-effectiveness and seamless integration. The system's non-intrusive design, reliance on affordable hardware, and modular architecture make it scalable for vehicular deployment. This study underscores the potential of EAR-based drowsiness detection systems to mitigate accidents caused by fatigue, offering a pragmatic solution for improving road safety in resource-constrained settings.