Drowsiness detection using Dlib: an overview

Drowsy driving represent a major factor for road accidents. Therefore, preventing these kind of accidents through drowsiness detection system is of great importance. Many systems exits to detect drowsiness signs that include yawning frequency, eye-blinking frequency, eye-gaze movement, head movement, and facial expressions. This paper provides an overview of driver drowsiness detection using Dlib that detects 68 facial landmarks. Dlib’s shape detector is utilized to accurately map and determine the coordinates of the facial landmarks. We focused on the landmarks of the eye and the mouth in order to calculate two important parameters that are the eye aspect ratio (EAR) and the mouth aspect ratio (MAR). These parameters gauge the degree of eye closure and the extent of mouth opening. An alert is triggered if the value of these parameters exceeds or falls below a predetermined threshold. In this review, several uses of Dlib face detection algorithm will be explored and compared in order to determine the optimal way to calculate the two parameters.