AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Image Inpainting via Conditional Texture and Structure Dual Generation

Xiefan Guo, Hongyu Yang, Di Huang

ICCV 2021 · Image generation › Autoregressive and latent generation · approach: Classical generative image synthesis · 367 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 structure priors via a structure stream (ablation)Texture-structure dual generation · Paris StreetView (mask 40-60%)LPIPS: 0.251 -> 0.226Improved
add structure priors via a structure stream (ablation)Texture-structure dual generation · Paris StreetView (mask 40-60%)PSNR: 22.22 -> 22.89Improved
use two-stream dual generation instead of multi-task single-stream (ablation)Texture-structure dual generation · Paris StreetView (mask 40-60%)LPIPS: 0.245 -> 0.226Improved
use two-stream dual generation instead of multi-task single-stream (ablation)Texture-structure dual generation · Paris StreetView (mask 40-60%)PSNR: 22.59 -> 22.89Improved
add Bi-GFF instead of concat+conv fusion (ablation)Texture-structure dual generation · Paris StreetView (mask 40-60%)LPIPS: 0.236 -> 0.226Improved
add Bi-GFF instead of concat+conv fusion (ablation)Texture-structure dual generation · Paris StreetView (mask 40-60%)SSIM: 0.567 -> 0.573Improved
add CFA module (ablation)Texture-structure dual generation · Paris StreetView (mask 40-60%)LPIPS: 0.243 -> 0.226Improved
add CFA module (ablation)Texture-structure dual generation · Paris StreetView (mask 40-60%)PSNR: 22.64 -> 22.89Improved
use multi-scale CFA instead of original contextual attention (ablation)Texture-structure dual generation · Paris StreetView (mask 40-60%)LPIPS: 0.240 -> 0.226Improved
use multi-scale CFA instead of original contextual attention (ablation)Texture-structure dual generation · Paris StreetView (mask 40-60%)SSIM: 0.566 -> 0.573Improved
use dual-generation model instead of RFRTexture-structure dual generation · Places2 (mask 40-60%)LPIPS: 0.239 -> 0.227Improved
use dual-generation model instead of MED (shared generator)Texture-structure dual generation · Places2 (mask 40-60%)LPIPS: 0.248 -> 0.227Improved

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