Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
MetaHTR: Towards Writer-Adaptive Handwritten Text Recognition
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Apply writer-adaptive MetaHTR to ASTER | ASTER · IAM | WRA (NL): 81.3 -> 89.2 | Helped |
| Apply writer-adaptive MetaHTR to SAR | SAR · IAM | WRA (NL): 84.4 -> 91.5 | Helped |
| Apply writer-adaptive MetaHTR to SCATTER | SCATTER · IAM | WRA (NL): 84.6 -> 91.6 | Helped |
| Apply writer-adaptive MetaHTR to ASTER | ASTER · RIMES | WRA (NL): 87.4 -> 93.4 | Helped |
| Apply writer-adaptive MetaHTR to SAR | SAR · RIMES | WRA (NL): 88.7 -> 93.7 | Helped |
| Apply writer-adaptive MetaHTR to SCATTER | SCATTER · RIMES | WRA (NL): 88.8 -> 93.9 | Helped |
| Add learnable character-instance loss weights instead of using mean cross-entropy (ablation) | MetaHTR · IAM | WRA (NL): 1.9%; 2.2%; 2.1% | Helped |
| Use character-instance gradients together with the mean-loss gradient to predict weights (ablation) | MetaHTR with ASTER · IAM | WRA (NL): 1.9% | Helped |
| Use layer-wise learnable learning rates instead of a fixed inner-loop learning rate (ablation) | MetaHTR · IAM | WRA (NL): 0.8%; 0.6%; 0.6% | Helped |
| Use one inference-time gradient update rather than more updates (ablation) | MetaHTR · Handwritten text recognition | WRA | Mixed |
| Increase the adaptation support-set size (ablation) | MetaHTR · Handwritten text recognition | WRA | Helped |
Closest papers by meaning
- MetaWriter: Personalized Handwritten Text Recognition Using Meta-Learned Prompt Tuning (CVPR 2025 · similarity 0.93)
- On the Generalization of Handwritten Text Recognition Models (CVPR 2025 · similarity 0.89)
- Muharaf: Manuscripts of Handwritten Arabic Dataset for Cursive Text Recognition (NeurIPS 2024 · similarity 0.84)
- General Detection-based Text Line Recognition (NeurIPS 2024 · similarity 0.84)
- Uni-MuMER: Unified Multi-Task Fine-Tuning of Vision-Language Model for Handwritten Mathematical Expression Recognition (NeurIPS 2025 · similarity 0.82)
- OCRBench v2: An Improved Benchmark for Evaluating Large Multimodal Models on Visual Text Localization and Reasoning (NeurIPS 2025 · 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.