Implementation of Real-Time Driver Drowsiness Detection

Drowsiness is observed to be a major contributor for road accidents, around 100,000 accidents per year.[1][2] Drowsiness is a silent killer as many are unaware of the risk entailing drowsiness unlike drunk driving as no system exist to measure drowsiness. Driver failing to concentrate on the road leads to a lesser reaction time and impaired steering behaviour. This paper proposes a low cost, ideal system to detect drowsiness unobtrusively and avoid accidents pertaining to the cause by taking preventive measures. Two primary inputs proposed in this work to accurately quantify the drowsiness are Camera Vision (CV) and Pulse rate sensor (PRS). CV quantifies the magnitude of eyelid closure and PRS obtained signal is used to detect pulse associated with the state of drowsiness. The proposed system uses Support Vector Machine (SVM) to classify the two states for drivers i.e. Drowsy State and Non-Drowsy State. When the driver exhibits alarming levels of drowsiness preventive measures are taken to avoid accident.