Video-based evaluation of driver's visual attention using smartphones

Abstract

In the recent period, video processing algorithms emerged as a valid solution to various problems, including the automation of some repetitive processes, for retrieving information from dynamic environments or the inspection of the physical state of human operators. On the other hand, with the latest advancements in the automotive industry, Advanced Driver Assistance Systems (ADASs) are becoming more and more common. It was just a matter of time until ADASs will rely heavily on the bio-cues processed from video steam. We present a system that obtains driver's head orientation, thus inferring the level of visual attention. The result of this research will be included in a Personal Navigation Assistant (PNA) based on dual camera mobile phones. We address the vision algorithms, the system architecture and the main implementation issues implied by the system.