From Gaze to Focus of Attention

  • Alex Waibel ,
  • Jie Yang ,
  • Michael Finke ,
  • Rainer Stiefelhagen

Identifying human gaze or eye-movement ultimately serves the purpose of identifying an individual’s focus of attention. The knowledge of a person’s object of interst helps us effectively communicate with other humans by allowing us to identify our conversants’ interests, state of mind, and/or intentions. In this paper we propose to track focus of attention of several participants in a meeting. Attention does not necessarily coincide with gaze, as it is a perceptual variable, as opposed to a physical one (eye or head positioning). Automatic tracking focus of attention is therefore achieved by modeling both, the persons head movements as well as the relative locations of probable targets of interst in a room. Over video sequences taken in a meeting situation, the focus of attention could be identified up to 98% of the time.