Software Centric Driver Fatigue Discernment System Using Computer Vision

In recent times, there has been a massive multiplication in the number of vehicles seen on the road. With this increase, there has, unfortunately, also been a significant rise in the daily recorded cases of road accidents. One of the topmost causes of road mishaps is tiredness of the driver driving the automobile. Distraction caused due to tiredness can be shunned by recognizing the moment when the driver is getting drowsy and alerting her/him. We aim to contribute to creating a system which determines the state of drowsiness and sends an alert to the driver so accidents can be prevented. We set up a real time face detection using a pre-trained facial landmark detector which marks different parts of the face by coordinates. Out of all the features detected, the eyes and mouth are fixated upon. We calculate the opening and shutting of eyes and the mouth as a descriptor of tiredness using different metrics. Then these metrics are measured against a threshold value for determination of fatigue. At last, if the detector labels the person as drowsy, an alarm goes off alerting the driver and deferring a mishap.