Driver Drowsiness Detection using Deep Learning

Drowsy driving is a contributing factor in majority of the car accidents occurring across the world. As a result, the most important approach for preventing these accidents is driver sleepiness detection, which can help us to reduce many road accidents. The goal of this article is to create a Sleepiness Detection System that detects whether a driver's eyes are closed for a few seconds, and then alerts the driver via an alarm. In this research, deep learning is used to suggest a new frame that classifies the driver's eye condition, i.e. open or closed. When a driver is determined to be sleepy, the proposed system sounds a beep on reaching a certain saturation point of the drowsiness measure. The proposed work is evaluated on a large part of MRL eye dataset consisting of 48000 images and it shows an accuracy of 86.05% using CNN model.