Lune

CVPR2023

AVFace: Towards Detailed Audio-Visual 4D Face Reconstruction

Aggelina Chatziagapi, Dimitris Samaras

DOI出版方

摘要

Figure 1 . Given a monocular talking face video and the corresponding speech segment, AVFace reconstructs the detailed 4D face geometry for any input speaker, and it is robust in cases when either modality is insufficient.