Evaluation of robust pose estimation methods within automotive environments

In order to support today's drivers in dealing with the huge amount of information in their environment, this paper gives an overview on algorithms to determine a major indicator for the driver's attentiveness and distraction level: the head's pose. This knowledge about the pose can be used subsequently to reduce the number of accidents on streets by activating warning strategies. After proposing those algorithms, they are evaluated and compared regarding the stability for automotive environments, such that the best system for pose estimation can be chosen. For this purpose evaluation criteria are proposed which cover those difficult situations in automotive environments.