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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| Train with the Fourier Converter to extract high-frequency information (ablation) | FDRNet · DocUNet | MS-SSIM: Not converge -> 0.32 | Improved |
| Add Mutual Transformation Loss (ablation) | FDRNet with CT and FC_tr · DocUNet | MS-SSIM: 0.32 -> 0.44 | Improved |
| Add Mutual Transformation Loss (ablation) | FDRNet with CT and FC_tr · DocUNet | LD: 34.16 -> 16.35 | Improved |
| Add Refinement Transformer (ablation) | FDRNet with CT, FC_tr, and MTL · DocUNet | MS-SSIM: 0.44 -> 0.50 | Improved |
| Add Refinement Transformer (ablation) | FDRNet with CT, FC_tr, and MTL · DocUNet | LD: 16.35 -> 9.43 | Improved |
| Add inference Fourier Converter (ablation) | FDRNet with CT, FC_tr, MTL, and RT · DocUNet | CER(%): 23.46 -> 16.96 | Improved |
| Use FDRNet rather than DocTr | Document dewarping · DocUNet | LD: 8.38 -> 9.43 | Hurt |
| Use FDRNet rather than DewarpNet | Document recognition · WarpDoc | CER(%): 45.82 -> 29.24 | Improved |
| Use FDRNet rather than GBSUM-Crop | Document dewarping · WarpDoc tightly cropped | MS-SSIM: 0.41 -> 0.46 | Improved |
| Set inference Fourier Converter β to 0.005 instead of 0.008 (ablation) | FDRNet · DocUNet | CER(%): 16.96 -> 17.84 | Hurt |
Closest papers by meaning
- DewarpNet: Single-Image Document Unwarping With Stacked 3D and 2D Regression Networks (ICCV 2019 · similarity 0.86)
- Revisiting Document Image Dewarping by Grid Regularization (CVPR 2022 · similarity 0.85)
- Improving the Learning Capability of Small-size Image Restoration Network by Deep Fourier Shifting (NeurIPS 2024 · similarity 0.83)
- DocRes: A Generalist Model Toward Unifying Document Image Restoration Tasks (CVPR 2024 · similarity 0.83)
- CoDe: An Explicit Content Decoupling Framework for Image Restoration (CVPR 2024 · similarity 0.82)
- Descanning: From Scanned to the Original Images with a Color Correction Diffusion Model (AAAI 2024 · similarity 0.82)
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.
