Driver fatigue surveillance via eye detection

In this study, driver alertness and fatigue-related surveillance were measured by image processing techniques for detecting the driver´s face and eyes in a frame. The image was acquired using an infrared-only camera that transforms human pupil into a distinct white circle; hence, the eyes are extracted more easily than those taken from a regular camera. The proposed model recorded eye closure measures, which are proven for the validation of fatigue. A multi-stage eye tracking process was also applied for ensuring robust, real-time eye movement. Meanwhile, a proposed warning module based on a back-propagation neural network employed as an artificial intelligence was used to train the program for adapting each individual. Finally, the proposed module attained a 97% success rate with high reliability at low cost.