The research aims to improve the accuracy of sleep detection in low light
conditions, as early detection of sleep is vital to improving safety and
performance, especially at work or on the go. Low illumination can significantly
reduce the accuracy of sleep detection, so the research aims to improve the
accurate detection of sleep in low light conditions. The methods used include
gamma correction on image processing to improve contrast and the use of eye
aspect ratio (EAR) to identify signs of drowsiness. Gamma correction helps
improve visibility of facial features in dark images, which is an important part of
advanced processing. The use of EAR is based on measuring the comparison
between open eyes and closed eyes, which is an important indicator of drowsiness.
The system can identify when a person starts showing signs of drowsiness, such as
blinking more often or closing his eyes longer than usual, by analyzing this ratio.
Research has shown that the application of this method increases the accuracy of
sleep detection in low light conditions significantly. Prior to using gamma
correction, the average accuracy was 21%, whereas after gamma corretion was
applied, the precision increased to an average of 91%. Thus, the results of this
study increased the detection precision by as much as 70% higher compared to
without gamma corrected in low light conditions.