Real time detection of driver drowsiness based on representation learning using deep neural networks

Drowsiness detection is a system that can detect a snoozing driver in order to prevent an accident.
Drowsiness is one of the main issues in road accidents. The fatality rate due to drowsiness
is higher. An accident involving driver drowsiness has a high fatality rate because the observation,
acknowledgement and vehicle control abilities reduce sharply while falling asleep. The
growing number of accident fatalities in world in recent years has become a problem of serious
concern for the society, so accidents must be prevented before they happen and this thing lies
with the driver. Accidents usually lead both economic as well as social loss to the society. If
accidents are prevented we can save many lives and along with that the environment is also
preserved. Preventing accidents caused by drowsiness requires a system for detecting sleepiness
in a driver. This work proposes a deep neural architecture for learning effective features
and detecting drowsiness for a given RGB input video of a driver. The architecture consists of
three deep networks for attaining global robustness to background and environmental variations
and learning local facial movements and head gestures important for reliable drowsiness
detection.

MPhil CS 2017-2018 INT Dr. Elizabeth Sherly