One important area of artificial intelligence (AI) research that directly affects healthcare, traffic safety, and human-machine interaction is the detection of drowsiness. One of the primary causes of traffic accidents, fatigue-related impairments are responsible for over 20% of serious collisions worldwide. Traditional detection methods that rely on steering changes, yawning occurrence, or eye aspect ratio limits have problems with reliability, sensitivity to lighting, and generalization. Convolutional, recurrent, and attention-driven neural networks are used in recent deep-learning (DL) techniques to effectively capture spatiotemporal features. This review summarizes current developments in AI-based drowsiness detection, including visual, physiological, and hybrid approaches. Architectures (CNN, LSTM, Transformer), evaluation standards, and research challenges are also described. A comparative analysis and outlook are offered for creating real-time, interpretable, and effective driver monitoring systems designed for smart transportation.
Ajudiya et al. (Wed,) studied this question.