AI Paper Atlas

Research map / Image generation / Editing and style transfer

On the Reconstruction of Face Images from Deep Face Templates

Guangcan Mai, Kai Cao, Pong Chi Yuen, Anil K. Jain

TPAMI 2018 · Image generation › Editing and style transfer · approach: Identity-preserving face manipulation · 241 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
use NbNet-B blocks instead of plain de-convolution (ablation)VGG-trained, perceptual loss · LFW type-IITAR @ FAR 0.1% (%): 45.63 -> 58.05Improved
use NbNet-B blocks instead of plain de-convolution (ablation)VGG-trained, pixel loss · LFW type-ITAR @ FAR 0.1% (%): 70.22 -> 89.52Improved
use NbNet-B blocks instead of plain de-convolution (ablation)MPIE-trained, pixel loss · LFW type-ITAR @ FAR 0.1% (%): 37.11 -> 67.86Improved
use NbNet-A blocks instead of plain de-convolution (ablation)MPIE-trained, perceptual loss · LFW type-ITAR @ FAR 0.1% (%): 85.34 -> 80.33Hurt
use NbNet-B blocks instead of plain de-convolution (ablation)MPIE-trained, perceptual loss · FRGC v2.0 type-ITAR @ FAR 0.1% (%): 55.22 -> 73.76Improved
add perceptual loss instead of pixel difference loss (ablation)NbNet-B trained on VGG-Face · LFW type-ITAR @ FAR 0.1% (%): 89.52 -> 94.37Improved
add perceptual loss instead of pixel difference loss (ablation)NbNet-B trained on Multi-PIE · LFW type-IITAR @ FAR 0.1% (%): 24.00 -> 37.30Improved
add DCGAN-generated image augmentation (ablation)NbNet-B trained on VGG-Face · LFW type-ITAR @ FAR 0.1% (%): 72.53 -> 89.52Improved
add DCGAN-generated image augmentation (ablation)NbNet-B trained on Multi-PIE · LFW type-ITAR @ FAR 0.1% (%): 34.87 -> 67.86Improved
add DCGAN-generated image augmentation (ablation)NbNet-B trained on Multi-PIE · color FERET type-I rank-1rank-1 (%): 63.88 -> 85.61Improved
train on mixed raw datasets instead of raw VGG-Face (ablation)NbNet-B pixel loss · LFW type-ITAR @ FAR 0.1% (%): 72.53 -> 71.62Hurt
use NbNet with DCGAN augmentation instead of RBF regressionNbNet-B perceptual loss · LFW type-ITAR @ FAR 0.1% (%): 19.76 -> 94.37Improved

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