An Efficient Approach for Detecting Driver Drowsiness Based on Deep Learning

Drowsy driving is one of the common causes road accidents resulting in injuries, even death, and significant economic losses to drivers, users, families, society. There have been many studies carried out an attempt detect drowsiness for alert systems. However, a majority focused on determining eyelid mouth movements, which revealed limitations detection. Besides, physiological measures-based may not be feasible practice because measuring devices are often available vehicles uncomfortable drivers. In this research, we therefore propose two efficient methods with three scenarios doze The former applies facial landmarks blinks yawns based appropriate thresholds each driver. latter uses deep learning techniques adaptive neural networks MobileNet-V2 ResNet-50V2. second method analyzes videos detects driver’s activities every frame learn all features automatically. We leverage advantage transfer technique train proposed our training dataset. This solves problem limited datasets, provides fast time, keeps networks. Experiments were conducted test effectiveness compared other methods. Empirical results demonstrate that using can achieve high accuracy 97%. study meaningful solutions prevent unfortunate automobile caused by drowsiness.