環境への自動適応を伴うアピアランスベース頭部姿勢推定

摘要

Head pose can be an important factor in inferring the focus of attention of humans, and thus can be used in a wide range of applications. For this reason, techniques for estimating head pose have been considered an important research task for decades. Although various image-based approaches have been proposed for estimating head pose (see 15) for a recent survey), one of the major remaining technical challenges is to deal with low resolution images. In some application scenarios like visual surveillance, it is often the case that head regions in input images are quite small. Small images contain limited information, thus it is still a challenging task to achieve accurate estimation results in such cases.