Driver drowsiness detection is a project built using Dlib and OpenCV with Python as a backend language.
The project includes direct working with the 68 facial landmark detector and also the face detector of the Dlib library. The 68 facial landmark detector is a robustly trained efficient detector which detects the points on the human face using which we determine whether the eyes are open or they are closed.The 68-landmark detector data (.dat) file can be found By clicking here
- As you can see the above screenshot where the landmarks aredetected using the detector.
- Now we are taking the ratio which is described as 'Sum of distances of vertical landmarks divided by twice the distance between horizontal landmarks'.
- Now this ratio is totally dependent on your system which you may configure accordingly for the thresholds of sleeping, drowsy, active.