AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Fourier Document Restoration for Robust Document Dewarping and Recognition

Chuhui Xue, Zichen Tian, Fangneng Zhan, Shijian Lu, Song Bai

CVPR 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Correspondence and causal representation learning · 39 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
Train with the Fourier Converter to extract high-frequency information (ablation)FDRNet · DocUNetMS-SSIM: Not converge -> 0.32Improved
Add Mutual Transformation Loss (ablation)FDRNet with CT and FC_tr · DocUNetMS-SSIM: 0.32 -> 0.44Improved
Add Mutual Transformation Loss (ablation)FDRNet with CT and FC_tr · DocUNetLD: 34.16 -> 16.35Improved
Add Refinement Transformer (ablation)FDRNet with CT, FC_tr, and MTL · DocUNetMS-SSIM: 0.44 -> 0.50Improved
Add Refinement Transformer (ablation)FDRNet with CT, FC_tr, and MTL · DocUNetLD: 16.35 -> 9.43Improved
Add inference Fourier Converter (ablation)FDRNet with CT, FC_tr, MTL, and RT · DocUNetCER(%): 23.46 -> 16.96Improved
Use FDRNet rather than DocTrDocument dewarping · DocUNetLD: 8.38 -> 9.43Hurt
Use FDRNet rather than DewarpNetDocument recognition · WarpDocCER(%): 45.82 -> 29.24Improved
Use FDRNet rather than GBSUM-CropDocument dewarping · WarpDoc tightly croppedMS-SSIM: 0.41 -> 0.46Improved
Set inference Fourier Converter β to 0.005 instead of 0.008 (ablation)FDRNet · DocUNetCER(%): 16.96 -> 17.84Hurt

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