Research map / Vision-language and multimodal / Multimodal LLMs
Table-MCR2TR: Merged-Cell-Aware Table Recognition via Reinforced Multimodal Language Models
Bangbang Zhou, Zhaoqing Zhu, Feiyu Gao, Hangdi Xing, Yadong Qu, Qi Zheng, Ming Yan, Hongtao Xie
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 MCT-350K data with TR-only SFT to base model (ablation) | Qwen2.5-VL-3B · CC-OCR Scan+Photo average | TEDS: 76.7 -> 83.5 | Improved |
| add MCT-350K data with TR-only SFT to base model (ablation) | Qwen2.5-VL-3B · CC-OCR Scan+Photo average | MCR F1: 46.4 -> 64.0 | Improved |
| add MCATR (3-task SFT plus M-CRL) on top of MCT-350K TR-only SFT (ablation) | Qwen2.5-VL-3B · CC-OCR Scan+Photo average | MCR F1: 64.0 -> 73.9 | Improved |
| add MCATR on top of MCT-350K TR-only SFT (ablation) | Qwen2.5-VL-3B · CC-OCR Scan+Photo average | TEDS: 83.5 -> 86.8 | Improved |
| add MCR F1 reward to alignment reward (ablation) | M-CRL · CC-OCR Scan+Photo average | MCR F1: 70.7 -> 73.9 | Improved |
| add TEDS reward to alignment and F1 rewards (ablation) | M-CRL · CC-OCR Scan+Photo average | TEDS: 85.7 -> 86.8 | Improved |
| add alignment reward to TEDS and F1 rewards (ablation) | M-CRL · CC-OCR Scan+Photo average | MCR F1: 73.2 -> 73.9 | Improved |
| add RL (M-CRL) after 3-task joint SFT (ablation) | Qwen2.5-VL-3B · CC-OCR Scan+Photo average | MCR F1: 67.9 -> 73.9 | Improved |
| use ground-truth merged-cell info instead of predicted MCR as input (ablation) | Table-MCR2TR · CC-OCR Scan | TEDS: 90.5 -> 93.9 (+3.4) | Improved |
| use ground-truth merged-cell info instead of predicted MCR as input (ablation) | Table-MCR2TR · CC-OCR Photo | TEDS: 83.0 -> 87.6 (+4.6) | Improved |
| use ground-truth merged-cell info instead of predicted MCR as input (ablation) | Table-MCR2TR · OmniDocBench tables | TEDS: 91.0 -> 93.6 (+2.6) | Improved |
| add M-CRL after 3-task SFT on a 1B model (ablation) | InternVL3-1B · CC-OCR Scan+Photo average | TEDS: 74.7 -> 80.2 | Improved |
Closest papers by meaning
- TRivia: Self-supervised Fine-tuning of Vision-Language Models for Table Recognition (CVPR 2026 · similarity 0.88)
- Grab What You Need: Rethinking Complex Table Structure Recognition with Flexible Components Deliberation (AAAI 2024 · similarity 0.87)
- TableDART: Dynamic Adaptive Multi-Modal Routing for Table Understanding (ICLR 2026 · similarity 0.87)
- Multimodal Tabular Reasoning with Privileged Structured Information (NeurIPS 2025 · similarity 0.87)
- TabPedia: Towards Comprehensive Visual Table Understanding with Concept Synergy (NeurIPS 2024 · similarity 0.86)
- TableMix: Enhancing Multimodal Table Reasoning in MLLMs from a Data-Centric Perspective (CVPR 2026 · similarity 0.86)
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.
