EMOVA: Emotion-driven neural volumetric avatar

Juheon Hwang, Byung-Gyu Kim, Taewan Kim, Heeseok Oh, Jiwoo Kang

Abstract: Highlights•A novel network, EMOVA, which uses two emotional stimuli from images and the corresponding voice to learn subtle differences in facial expression.•The attention-based fusion method to efficiently combine the information of images and voice, resulting in the high-quality reconstruction of faces.•Through exhaustive evaluations, EMOVA is demonstrated to reconstruct significantly more realistic and expressive faces than state-of-the-art methods.