Embodied conversational agents : computing and rendering realistic gaze patterns
Résumé
We describe here our efforts for modeling multimodal signals exchanged by interlocutors when interacting face-to-face. This data is then used to control embodied conversational agents able to engage into a realistic face-to-face interaction with human partners. This paper focuses on the generation and rendering of realistic gaze patterns. The problems encountered and solutions proposed claim for a stronger coupling between research fields such as audiovisual signal processing, linguistics and psychosocial sciences for the sake of efficient and realistic human-computer interaction.
Origine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...