AI Paper Atlas

Research map / Image generation / Image restoration

High-Resolution Image Harmonization via Collaborative Dual Transformations

Wenyan Cong, Xinhao Tao, Li Niu, Jing Liang, Xuesong Gao, Qihao Sun, Liqing Zhang

CVPR 2022 · Image generation › Image restoration · approach: Image harmonization and color transformation · 90 citations (OpenAlex)

Check 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
Combine low-resolution pixel-to-pixel transformation, high-resolution RGB-to-RGB transformation, and refinementCDTNet · HAdobe5k 1024×1024PSNR: 37.65 -> 38.77Improved
Add RGB-to-RGB transformation to the pixel-only system (ablation)CDTNet · HAdobe5k 1024×1024PSNR: 29.41 -> 38.77Improved
Add pixel-to-pixel transformation and refinement to RGB-only transformation (ablation)CDTNet · HAdobe5k 1024×1024PSNR: 37.65 -> 38.77Improved
Share encoder E between pixel harmonization and RGB weight prediction (ablation)CDTNet color mapping module · HAdobe5k 1024×1024PSNR: 36.86 -> 37.65Improved
Pool foreground and background features separately before predicting LUT coefficients (ablation)CDTNet color mapping module · HAdobe5k 1024×1024PSNR: 37.31 -> 37.65Improved
Include the high-resolution RGB transformation result in refinement (ablation)CDTNet refinement module · HAdobe5k 1024×1024PSNR: 37.29 -> 38.77Improved
Include the upsampled pixel-to-pixel result in refinement (ablation)CDTNet refinement module · HAdobe5k 1024×1024PSNR: 37.78 -> 38.77Improved
Include the high-resolution mask in refinement (ablation)CDTNet refinement module · HAdobe5k 1024×1024PSNR: 37.33 -> 38.77Improved
Include the upsampled decoder feature in refinement (ablation)CDTNet refinement module · HAdobe5k 1024×1024PSNR: 37.64 -> 38.77Improved
Use a blending layer in refinement (ablation)CDTNet refinement module · HAdobe5k 1024×1024PSNR: 28.13 -> 38.77Improved
Use deep RGB-to-RGB transformation only at inference instead of the full collaborative method (ablation)CDTNet-256 · HAdobe5k 1024×1024PSNR: 38.77 -> 37.65Hurt
Use deep RGB-to-RGB transformation only at inference instead of the full collaborative method (ablation)CDTNet-256 · HAdobe5k 2048×2048PSNR: 37.66 -> 37.28Hurt

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