AI Paper Atlas

Research map / 3D vision / Face and shape

Cross-Modal Perceptionist: Can Face Geometry Be Gleaned From Voices?

Cho-Ying Wu, Chin-Cheng Hsu, Ulrich Neumann

CVPR 2022 · 3D vision › Face and shape · approach: Speech-driven 3D facial animation · 16 citations (OpenAlex)

Check your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
jointly train speech-to-image and image-to-3D prediction with voice information rather than cascade separately pretrained modulesCMP-supervised · Voxceleb-3D testMean ARE: 0.0302 -> 0.0241Improved
jointly train speech-to-image and image-to-3D prediction with voice information rather than cascade separately pretrained modulesCMP-unsupervised · Voxceleb-3D testMean ARE: 0.0302 -> 0.0251Improved
learn face geometry from voice using the supervised paired-voice/mesh frameworkCMP-supervised · Voxceleb-3D testARE ER: 0.0311 -> 0.0152Improved
learn face geometry from voice using the supervised paired-voice/mesh frameworkCMP-supervised · Voxceleb-3D testARE FR: 0.0173 -> 0.0186Hurt
learn face geometry from voice using the unsupervised joint frameworkCMP-unsupervised · Voxceleb-3D testARE ER: 0.0311 -> 0.0181Improved
learn face geometry from voice using the unsupervised joint frameworkCMP-unsupervised · Voxceleb-3D testARE MR: 0.0173 -> 0.0174Hurt
use RKD or PKT instead of vanilla KD (ablation)CMP-unsupervised · Voxceleb-3D unsupervisedMean ARE: 0.0298 -> 0.0254Improved
use attention-map similarity KD instead of vanilla KD (ablation)CMP-unsupervised · Voxceleb-3D unsupervisedMean ARE: 0.0298 -> 0.0304Hurt
use SP KD instead of vanilla KD (ablation)CMP-unsupervised · Voxceleb-3D unsupervisedMean ARE: 0.0298 -> 0.0262Improved
use Correlation KD instead of vanilla KD (ablation)CMP-unsupervised · Voxceleb-3D unsupervisedMean ARE: 0.0298 -> 0.0263Improved
use CRD instead of vanilla KD (ablation)CMP-unsupervised · Voxceleb-3D unsupervisedMean ARE: 0.0298 -> 0.0255Improved
use VID instead of vanilla KD (ablation)CMP-unsupervised · Voxceleb-3D unsupervisedMean ARE: 0.0298 -> 0.0259Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy