Expressive Virtual Human: Impact of expressive wrinkles and pupillary size on emotion recognition

Abstract

Improving the expressiveness of virtual humans is essential for qualitative interactions and development of an emotional bond. It is certainly indicated for all applications using the user's cognitive processes, such as applications dedicated to training or health. Our study aims to contribute to the design of an expressive virtual human, by identifying and adapting visual factors promoting transcription of emotions. In this paper, we investigate the effect of expressive wrinkles and variation of pupil size. We propose to compare the recognition of basic emotions on a real human and on an expressive virtual human. The virtual human was subject to two different factors: expressive wrinkles and/or pupil size. Our results indicate that emotion recognition rates on the virtual agent are high. Moreover, expressive wrinkles affect emotion recognition. The effect of pupillary size is less significant. However, both are recommended to design an expressive virtual human.