Head pose classification for passenger with CNN

Abstract

The utilization of autonomous driving technologies has grown over the past few decades, but public perception has brought into question its effectiveness. With the potential to revolutionize traffic safety, its application is crucial. Apprehensiveness toward automated vehicles can be targeted by developing a reliable intelligence system tailored to the passenger's judgement of said system. The goal of this study is to examine passenger behavior in relation to the driver to develop applicable control structures that are able to react to responses from passengers. By employing machine learning techniques to identify the head position of passengers from video, this paper investigates a method for using remotely sensed passenger head position to understand the passengers' judgement of the driver. The results of this study reveal that passenger behavior can be assessed using remote head position sensing and their behavior can be used as a control structure in vehicles. We found that passengers kept their head straight majority of the drive which we consider as confirmation of the potential for future uses.