Comparative Effectiveness of Deep Learning Approaches for Drowsiness Detection in a Demographically Diverse Cohort

Abstract

Modern lifestyles often demand individuals to balance multiple responsibilities, leading to inadequate sleep and compromised alertness. This can impact various aspects of life, from productivity at work to safe operation of vehicles. This work acknowledges that drowsiness is not confined to particular situations such as driving; it can manifest during work, studying, or any activity demanding prolonged focus among all aged individuals. The proposed Artificial Intelligence Based Sleep Harness System (ASHS), featuring Deep Learning models, is not only a technological support but also a social solution aimed at caring for society's well-being. By integrating drowsiness detection models with remedial measures, contributes significantly to public safety and health across all age groups. ASHS utilizes deep learning methods such as Convolutional-Neural-Network (CNN), Faster-Region-Based CNN (FR-CNN), and Recurrent-Neural-Networks (RNN) to detect and address drowsiness. This research work monitors facial expressions, eye movements, and head tracking to comprehensively assess drowsiness. When drowsiness is detected, real-time alerts are issued, providing immediate corrective measures. Beyond individual well-being of all aged ones, this ASHS has a societal impact by enhancing safety in transportation, productivity in the workplace, and improved performance in educational settings. Additionally, it aids in elderly care by monitoring sleep quality and health, ensuring timely support.