Research map / Image generation / Image restoration
DocScanner: Robust Document Image Rectification with Progressive Learning
Hao Feng, Wengang Zhou, Jiajun Deng, Qi Chuan Tian, Houqiang Li
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add document localization module (ablation) | DocScanner-B · DocUNet Benchmark | LD: 9.22 -> 7.62 | Improved |
| add document localization module (ablation) | DocScanner-B · DocUNet Benchmark | Li-D: 2.23 -> 1.88 | Improved |
| add document localization module (ablation) | DocScanner-B · DocUNet Benchmark | CER: 0.1734 -> 0.1788 | No clear effect |
| replace bilinear with learnable upsampling of flow residual (ablation) | DocScanner-B · DocUNet Benchmark | LD: 8.03 -> 7.62 | Improved |
| use ConvLSTM instead of ConvGRU (ablation) | DocScanner-B · DocUNet Benchmark | LD: 7.62 -> 7.92 | No clear effect |
| untie weights across iterations (ablation) | DocScanner-B · DocUNet Benchmark | LD: 7.62 -> 7.52 | No clear effect |
| untie weights across iterations (ablation) | DocScanner-B · DocUNet Benchmark | Parameters (M): 5.2 -> 38.7 | Hurt |
| add circle-consistency loss (ablation) | DocScanner-B · DocUNet Benchmark | LD: 7.99 -> 7.62 | Improved |
| add circle-consistency loss (ablation) | DocScanner-B · DocUNet Benchmark | Li-D: 1.99 -> 1.88 | Improved |
| add warped feature to flow updater input (ablation) | DocScanner-B · DocUNet Benchmark | LD: 9.09 -> 9.03 | No clear effect |
| add flow features to flow updater input (ablation) | DocScanner-B · DocUNet Benchmark | Li-D: 2.70 -> 1.97 | Improved |
| fuse warped and flow features in rectified feature generator (ablation) | DocScanner-B · DocUNet Benchmark | LD: 8.01 -> 7.62 | Improved |
Closest papers by meaning
- DocRes: A Generalist Model Toward Unifying Document Image Restoration Tasks (CVPR 2024 · similarity 0.85)
- DocFormerv2: Local Features for Document Understanding (AAAI 2024 · similarity 0.85)
- Towards Real-World Document Parsing via Realistic Scene Synthesis and Document-Aware Training (CVPR 2026 · similarity 0.82)
- Boosting Document Parsing Efficiency and Performance with Coarse-to-Fine Visual Processing (CVPR 2026 · similarity 0.82)
- Descanning: From Scanned to the Original Images with a Color Correction Diffusion Model (AAAI 2024 · similarity 0.82)
- DocPTBench: Benchmarking End-to-End Photographed Document Parsing and Translation (NeurIPS 2026 · 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.
