AI Paper Atlas

Research map / 3D vision / Face and shape

A Decoupled 3D Facial Shape Model by Adversarial Training

Victoria Fernandez Abrevaya, Adnane Boukhayma, Stefanie Wuhrer, Edmond Boyer

ICCV 2019 · 3D vision › Face and shape · approach: 3D morphable face model regression · 44 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 mesh regularization (ablation)Decoupled 3D face GAN · Decoupled 3D face model evaluationspecificity (mm): 3.61 -> 2.01Improved
add mesh regularization (ablation)Decoupled 3D face GAN · Decoupled 3D face model evaluationglobal diversity (mm): 1.41 -> 4.74Improved
add identity classification (ablation)Decoupled 3D face GAN · Decoupled 3D face model evaluationidentity decoupling (%): 7.8 -> 98.6Improved
add expression classification (ablation)Decoupled 3D face GAN · Decoupled 3D face model evaluationexpression decoupling (%): 42.8 -> 89.7Improved
add feature loss (ablation)Decoupled 3D face GAN · Decoupled 3D face model evaluationexpression decoupling (%): 80.3 -> 89.7Improved
add feature loss (ablation)Decoupled 3D face GAN · Decoupled 3D face model evaluationglobal diversity (mm): 4.47 -> 4.74Improved
add feature loss (ablation)Decoupled 3D face GAN · Decoupled 3D face model evaluationidentity decoupling (%): 96.0 -> 98.6Hurt
use adversarial 3D-2D model instead of 3DMMDecoupled 3D face GAN · Decoupled 3D face model evaluationexpression decoupling (%): 65.6 -> 89.7Improved
use adversarial 3D-2D model instead of MAE autoencoderDecoupled 3D face GAN · Decoupled 3D face model evaluationexpression decoupling (%): 53.3 -> 89.7Improved
use adversarial 3D-2D model instead of CoMA/DeepFLAMEDecoupled 3D face GAN · Decoupled 3D face model evaluationglobal diversity (mm): 3.38 -> 4.74Improved
use adversarial 3D-2D model instead of 3DMMDecoupled 3D face GAN · Decoupled 3D face model evaluationidentity decoupling (%): 99.6 -> 98.6Hurt
use adversarial 3D-2D model instead of CoMADecoupled 3D face GAN · Sparse landmark reconstruction (lambda=0)RMSE with no regularization (mm): 3.05 -> 2.62Improved

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