Advanced detection techniques for driver drowsiness: a comprehensive review of machine learning, deep learning, and physiological approaches

Road mishaps, a global concern, see driver fatigue contributing to roughly 40% of accidents in India. Detecting drowsiness early is pivotal in countering this threat. Despite prior research in this domain, many investigations fall short in utilizing extensive, varied datasets or analyzing real-time video streams, both essential for practical implementations. Our research addresses these shortcomings by using a range of classifiers, such as Convolutional Neural Networks (CNN), Support Vector Machines (SVM), Hidden Markov Models (HMM), Fuzzy Logic, and several sensors, to assess eye movements and other physiological signs. An in-depth evaluation indicates that CNNs are superior, especially when handling expansive, diverse image sets. This paper provides a comprehensive overview of current models, their pros and cons, classification techniques, labeling of drowsiness, and detection approaches, serving as a valuable reference for researchers and professionals aiming to create enhanced systems for real-time driver fatigue detection.