AI Paper Atlas

Research map / Image generation / Editing and style transfer

ChunkyGAN: Real Image Inversion via Segments

Adéla Šubrtová, David Futschik, Jan Čech, Michal Lukáč, Eli Shechtman, Daniel Sýkora

ECCV 2022 · Image generation › Editing and style transfer · approach: GAN inversion for editing · 4 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
optimize separate latent codes per segment instead of one global codeW+ projection · CelebA 100 images, projection fidelityLPIPS: 0.3675 -> 0.3194Improved
optimize separate latent codes per segment instead of one global codeW projection · CelebA 100 images, projection fidelityLPIPS: 0.4190 -> 0.3697Improved
optimize separate latent codes per segment instead of one global codeS-space projection · CelebA 100 images, projection fidelityLPIPS: 0.3577 -> 0.3572No clear effect
optimize separate latent codes per segmentW+ projection · CelebA 100 images, projection fidelityL2 loss: 0.0436 -> 0.0207Improved
optimize separate latent codes per segmentW+ projection · CelebA 100 images, projection fidelityidentity loss: 0.1195 -> 0.0937Improved
add latent code regularization lambda_reg=1 (ablation)Ours in W+ · CelebA 100 images, projection fidelityLPIPS: 0.3194 -> 0.3330Hurt
add latent code regularization lambda_reg=1 (ablation)Ours in W+ · CelebA 100 images, projection fidelityidentity loss: 0.0937 -> 0.0894Improved
add latent code regularization lambda_reg=1 (ablation)Ours in W+ · CelebA editing, gender, projected vs editedidentity loss: 0.298 -> 0.126Improved
add latent code regularization lambda_reg=1 (ablation)Ours in W+ · CelebA editing, beard, projected vs editedidentity loss: 0.312 -> 0.091Improved
add latent code regularization lambda_reg=1 (ablation)Ours in W+ · CelebA editing, age, projected vs editedidentity loss: 0.151 -> 0.069Improved
use segment-wise codes without regularization for editingOurs in W+ · CelebA editing, gender, projected vs editedidentity loss: 0.209 -> 0.298Hurt

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