Development and Validation of Basic Virtual Human Facial Emotion Expressions.

This paper introduces the design process of facial expressions on virtual humans to play basic emotions. The design of the emotions is grounded on the Facial Action Coding System that enables describing facial expressions based on Action Units. All the tools employed to attain the final human avatar expressions are detailed. Then, an experiment with healthy volunteers is carried out to validate the designed virtual human facial emotions. As result, we obtained that the faces are correctly interpreted by healthy people with an accuracy of \(83.56\%\). Thus, as recognition works quite well with this small sample of healthy people, this paper is a first step towards validating and enhancing the avatar characters generated, experimenting with a sufficient number of healthy persons, and, then, designing therapies based on human avatars to enhance facial affect recognition in patients with deficits in facial affect recognition.