AI Paper Atlas

Research map / 3D vision / Face and shape

Expression-Preserving Face Frontalization Improves Visually Assisted Speech Processing

Zhiqi Kang, Mostafa Sadeghi, Radu Horaud, Xavier Alameda-Pineda

IJCV 2023 · 3D vision › Face and shape · approach: Speech-driven 3D facial animation · 7 citations (OpenAlex)

Read the paperCheck 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
add dynamic Kalman-style shape update to robust frontalizationRFF · OuluVS2 frontalizationZNCC: 0.831 -> 0.839Improved
add robust RFF frontalization to the lip input (ablation)Res-AV-CVAE · MEAD AVSE SNR 0 dBSTOI: 0.67 -> 0.71Improved
add dynamic frontalization (DFF) to the lip input (ablation)Res-AV-CVAE · MEAD AVSE SNR 0 dBSTOI: 0.67 -> 0.73Improved
add dynamic frontalization (DFF) to the lip input (ablation)Res-AV-CVAE · MEAD AVSE SNR 0 dBPESQ: 1.99 -> 2.20Improved
add dynamic frontalization (DFF) to the lip input (ablation)Res-AV-CVAE · MEAD AVSE SNR 0 dBSI-SDR (dB): 4.68 -> 5.87Improved
use DFF instead of RFFRes-AV-CVAE · MEAD AVSE SNR 0 dBSI-SDR (dB): 5.24 -> 5.87Improved
use ST-GAN frontalization instead of noneRes-AV-CVAE · MEAD AVSE SNR 0 dBSTOI: 0.67 -> 0.68No clear effect
use DA-GAN frontalization instead of noneRes-AV-CVAE · MEAD AVSE SNR 0 dBPESQ: 1.99 -> 1.87Hurt
replace fully connected visual encoder with pretrained ResNet lip-reading backboneAV-CVAE · MEAD AVSESTOI/PESQImproved
fine-tune lip reader with proposed frontalization instead of GAN frontalizationTCN lip-reading model · LRW 500-IWRWCS (%): 84 -> 88Improved
fine-tune lip reader with GAN frontalizationTCN lip-reading model · LRW 500-IWRWCS (%): 87 -> 84Hurt
fine-tune lip reader with proposed frontalizationTCN lip-reading model · LRW 500-IWRWCS (%): 87 -> 88Improved

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