Drowsiness and fatigue detection using multi-feature fusion

Road accidents are one of the major concerns these days. Most of these accidents happen either due to inattentive driving or fatigue and drowsiness. With the advancement in computer vision and image processing techniques, it has become possible to rely on artificial intelligence to help reduce the threat to human life by these accidents. This can be achieved by installing built-in driver assistance systems in automobiles. This work is dedicated to developing a real-time and fast-acting drowsiness and fatigue detection model. The proposed model uses facial landmarks that detect and tracks the movement of the eyes and mouth to predict the state of the driver. Hence this model can save a number of valuable lives and prevent accidents from happening by activating an alarm whenever a driver feels dizzy. This model uses hybrid parameters which increases the reliability of the system. Also, the proposed model is low-cost and can be included in automobiles as an additional feature with a slight increase in cost. The proposed model can detect fatigue and drowsiness at 65 fps and has a high accuracy of 98.16%.