Animating virtual signers : the issue of gestural anonymization

Félix Bigand (1, 2) , E. Prigent (1) , Annelies Braffort (1, 2)

1 LISN - Laboratoire Interdisciplinaire des Sciences du Numérique
2 CNRS - Centre National de la Recherche Scientifique

This paper presents an ongoing PhD research project on visualperception and motion analysis applied to virtual signers (virtualagents used for Sign Language interaction).Virtual signers (or signing avatars) play an important role inthe accesibility of information in sign languages. They have beendeveloped notably for their capability to anonymize shape and ap-pearance of the content producer. While motion capture provideshuman-like, realistic and comprehensible signing animations, italso arises the question of anonymity. Human body movementscontain important information about a personâs identity, genderor emotional state. In the present work, we want to address theproblem of gestural identity in the context of animated agents inFrench Sign Language. The ability to identify a person from sign-ing motion will be assessed through psychophysical experiments,using point-light displays. We will then develop a computationalframework in order to investigate which features are critical forperson identification and to control them over the virtual agent.

Mots clés: Identity Recognition; Perception; Sign Language; Virtual Signer; Motion Capture

Domaines: Informatique [cs]

Félix Bigand, E. Prigent, Annelies Braffort. Animating virtual signers : the issue of gestural anonymization. 19th International Conference on Intelligent Virtual Agents (ACM IVA 2019), ACM, Jul 2019, Paris, France. pp.252-255, ⟨10.1145/3308532.3329410⟩. ⟨hal-02400928⟩