Table 3.

Some of the image-based measures.

Features	Description
Blink frequency [30]	The number of times an eye closes over a specific period of time.
Maximum closure duration of the eyes [30]	The maximum time the eye was closed. However, it can be risky to delay detecting an extended eye closure that indicates a drowsy driver.
Percentage of eyelid closure (PERCLOS) [31]	The percentage of time (per minute) in which the eye is 80% closed or more.
Eye aspect ratio (EAR) [32]	EAR reflects the eye’s openness degree. The EAR value drops down to zero when the eyes are closed. On the other hand, it remains approximately constant when the eye is open. Thus, the EAR detects the eye closure at that time.
Yawning frequency [33]	The number of times the mouth opens over a specific period of time.
Head pose [34]	Is a figure that describes the driver’s head movements. It is determined by counting the video segments that show a large deviation of three Euler angles of head poses from their regular positions. These three angles are nodding, shaking, and tilting.