@romanzadyrako We use real-time analytics. On the first start we request a user to calibrate the device so we can detect user's head normal driving position. Later on we will adjust the algorithm to learn user's head position automatically, so the "calibration" would not be needed, but at this time there is a simple "touch of a button" calibration. Then we can monitor and score certain events, such as eye blinking frequency, duration of eyes closing, relative head rotation, yawning and so on. Depending on event or overall scoring, we produce different alerts. It may be more or less mild reminder like "keep your eyes on the road" if we detect a user not looking ahead for a prolonged period of time, or rather stiff warning with loud and screechy beep: "Danger! Danger! Wake up!" if we detect that a driver closes their eyes. So in a nutshell, we constantly monitor the user's face and take an appropriate action as situation requires: from alerting the user to keep off their smartphone to waking up the driver to recommending to take a rest