Visual Focus of Attention Recognition in the Ambient Kitchen
This paper presents a model for visual focus of attention recognition in the Ambient Kitchen, a pervasive computing prototyping environment. The kitchen is equipped with several blended displays on one wall and users may use information presented on these displays from multiple locations. Our goal is to recognize which display the user is looking at so that the environment can adjust the display content accordingly. We propose a dynamic Bayesian network model to infer the focus of attention, which models the relation between multiple foci of attention, multiple user locations and faces captured by the multiple cameras in the environment. Head pose is not explicitly computed but measured by a similarity vector which represents the likelihoods of multiple face clusters. Video data are collected in the Ambient Kitchen environment and experimental results demonstrate the effectiveness of our model.
Unable to display preview. Download preview PDF.