Deep Learning based Driver Drowsiness Detection

Abstract

Driver drowsiness and distraction are now widely recognized as important contributors to deadly road accidents around the world. This typically occurs when a driver has not slept enough, but it can also occur as a result of untreated sleep problems, drugs, alcohol consumption, or shift employment. As a result, driver monitoring and identification are becoming increasingly important features of car safety systems. Head position, gaze direction, yawning, and eye state analysis are among the essential aspects. This research proposes a driver drowsiness detection system that uses eye blink, mouth opening and closing counts to detect drowsiness. When the driver's eyes are closed for an extended period of time, an alert sound is generated to notify him. Furthermore, the vehicle's owner is notified by e-mail if the driver is observed to be napping off more than a few times, in order to certify that the driver is taking certain steps to avoid falling asleep. The output of the system proposed in the paper on Deep learning technology of Dlib which uses CNN (Convolutional Neural Network) as its base algorithm for accurate detection, OpenCV, and Raspberry Pi environments with a mounted camera for the same, show that system achieves good result when it comes to drowsiness detection, reducing the overall number of accidents on the streets. For Realtime video input, the proposed method had a 96% of accuracy.