FEER: Non-intrusive facial expression and emotional recognition for driver's vigilance monitoring

Each year, drivers' loss of vigilance is chasing human lives in almost 25% of road accidents. In this paper, we present a non-intrusive approach that relies on facial expression detection. Face features, such as eyes and mouth, are extracted and quickly analyzed, using an integrated camera with an onboard processor. The closure rate and frequency of the eyes is then combined with the rate and frequency of yawning in a weighted combination to compute a decision map. Based on that decision, actions can range from a simple warning, to a severe warning, and sometimes taking control of the vehicle, such as automatic braking or deceleration. The proposed approach proved to be fast and accurate in terms of sleepiness detection with a very low rate of false positives.