Real-Time Driver Drowsiness Detection Using Mobile Net for Road Safety Enhancement

Research shows that fatigue ranks within the three main road accident causes worldwide. This paper develops a driver sleep detection system in real time with MobileNet as the light weight convolutional neural network architecture built specifically for indoor and mobile environments. Real-time system facial scans seek driver indicators of drowsiness which include lengthy eye closures and slow yawn activity leading to sleepiness. MobileNet operates at a low cost which makes it both rapid and functional for vehicle implementation without accuracy compromises. The system helps prevent driver fatigue which leads to better vehicle safety because drivers receive instant feedback through this alert model. This solution presents several advantages such as affordability as well as mobility through mobile devices which makes it ideal for current vehicle safety requirements. Fatigue signs to avoid accidents. When drowsiness accelerates the slowdown of reaction times but increases alertness it causes higher risks of vehicle crashes. System implementations that depend on physical signals need drivers to maintain close positions with the equipment which results in driver distraction. We evaluate the driver eye movement behavior to measure driver behavior through imaging-based dataset measurements in this paper. MobileNet eliminates optical images successfully at a low cost thus speeding up the visual fatigue relief process. MobileNet model demonstrated a performance of 94 % in the evaluation results. The system requires 0.00829 seconds to recognize tiredness. The system delivers exceptional real-time functionality thus making it an extraordinary example. Better performance was achieved through the implementation of hyperparameter optimization in connection with k fold validation. The study matters greatly because its implementation would assist in protecting lives by presenting an effective strategy to stop road fatalities arising from unpredictable seizures.