Section: New Results
Distant Human Representation for Computer-mediated Communications: Mixing Video and Avatar
Participants : Christophe Chaillou, Wang Haibo, Ding Li.
The research on distant human representation in computer-mediated communications has considerably increased in the last decade. This paper, written in collaboration with Casia & Tsinghua, proposes a novel human representation method. In the proposal, two non-verbal communication channels, facial and gesture expressions, as the crucial supplements to verbal cues, are executed by mixing video and avatar. In more detail, the facial and gesture information are first extracted from live image sequences. Then the extracted facial information, in the form of texture image, is mapped on an avatar which represents a human user. Meanwhile, the extracted gesture information is used to drive the avatar by the featurematching techniques. Our experimental results  show that the proposed method are sufficient to distantly represent human user.
We also propose a gaze estimation method that considers both eye gaze and coordinated head orientations  .