Research map / Vision-language and multimodal / Human and embodied modeling
An End-to-End OCR Text Re-organization Sequence Learning for Rich-text Detail Image Comprehension
Liangcheng Li, Feiyu Gao, Jiajun Bu, Yongpan Wang, Zhi Yu, Qi Zheng
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 Sinkhorn normalization to the pointer decoder (ablation) | GCN-PN · DI test set | total order accuracy: 0.79 -> 0.86 | Improved |
| add Sinkhorn normalization to the pointer decoder (ablation) | GCN-PN · DI test set | BLEU-4: 0.71 -> 0.74 | Improved |
| add Sinkhorn normalization to the pointer decoder (ablation) | GCN-PN · DI test set | BLEU-2: 0.90 -> 0.92 | Improved |
| replace MLP pairwise order prediction with GCN encoder and pointer decoder | POS-MLP · DI test set | total order accuracy: 0.75 -> 0.79 | Improved |
| replace position-only heuristics with learned image-plus-geometry model | POS-Hier · DI test set | total order accuracy: 0.70 -> 0.86 | Improved |
| replace greedy nearest-block ordering with GCN-PN-Sinkhorn | POS-Greedy · DI test set | total order accuracy: 0.41 -> 0.86 | Improved |
| POS-Hier versus GCN-PN on local versus longer sub-sequences | POS-Hier · DI test set | BLEU-2 vs BLEU-4: BLEU-2 0.89, BLEU-4 0.66 | Mixed |
Closest papers by meaning
- XYLayoutLM: Towards Layout-Aware Multimodal Networks for Visually-Rich Document Understanding (CVPR 2022 · similarity 0.85)
- Joint Visual Semantic Reasoning: Multi-Stage Decoder for Text Recognition (ICCV 2021 · similarity 0.85)
- PreSTU: Pre-Training for Scene-Text Understanding (ICCV 2023 · similarity 0.85)
- PP-OCRv5: A Specialized 5M-Parameter Model Rivaling Billion-Parameter Vision-Language Models on OCR Tasks (CVPR 2026 · similarity 0.84)
- Hierarchical Visual Feature Aggregation for OCR-Free Document Understanding (NeurIPS 2024 · similarity 0.84)
- DocLayLLM: An Efficient Multi-modal Extension of Large Language Models for Text-rich Document Understanding (CVPR 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.
