Deep-Learning-Based Drowsiness Detection Using Eye Closure and Yawn Analysis

The dataset used for the drowsiness detection system is a curated collection of real-time facial images and video frames designed to identify signs of fatigue. It includes labeled data for eye states (open and closed) and mouth states (yawning and not yawning) captured under varying lighting conditions, facial orientations, and demographic diversity. The dataset features over 10,000 annotated frames, with balanced representation across classes to ensure robust model performance. Preprocessing involved normalization, augmentation, and resizing for deep learning compatibility. This dataset facilitated training and validation of models with high accuracy, enabling real-time detection of drowsiness through transfer learning and computer vision techniques.