---
title: "MessToClean — paper by Tuo et al. (2026) — summary and ideas"
description: "This makes it practical to build auditable question banks from messy real-world photos. 29.7 points. 50 references."
authors: ["Jiayi Tuo", "Cheng Tang", "Zihan Wang", "Chenyue Zhou", "Yao Li", "Yanbiao Ma", "Chao Wang", "Wei Dai", "Mingxuan Wang", "Shitong Qin", "Ziwei Zhao"]
year: 2026
venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"
source: "https://aclanthology.org/2026.acl-long.1871/"
doi: "10.18653/v1/2026.acl-long.1871"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/tuo-2026-messtoclean-evidence-grounded-structure-preserving/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/tuo-2026-messtoclean-evidence-grounded-structure-preserving/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/36/) / MessToClean: Evidence-Grounded Structure-Preserving Reconstruction for Real-World Degraded Exam Paper Images

# MessToClean: Evidence-Grounded Structure-Preserving Reconstruction for Real-World Degraded Exam Paper Images

**Authors:** Jiayi Tuo, [Cheng Tang](https://md.research.elamur.ai/free/en/authors/cheng-tang/), [Zihan Wang](https://md.research.elamur.ai/free/en/authors/zihan-wang/), Chenyue Zhou, Yao Li, Yanbiao Ma, [Chao Wang](https://md.research.elamur.ai/free/en/authors/chao-wang/), [Wei Dai](https://md.research.elamur.ai/free/en/authors/wei-dai/), [Mingxuan Wang](https://md.research.elamur.ai/free/en/authors/mingxuan-wang/), Shitong Qin and Ziwei Zhao · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.acl-long.1871/) · [PDF](https://aclanthology.org/2026.acl-long.1871.pdf) · [DOI 10.18653/v1/2026.acl-long.1871](https://doi.org/10.18653/v1/2026.acl-long.1871) · [OpenAlex](https://openalex.org/W7166852091)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Exam sheets collected by education platforms are usually phone photos. They tend to be skewed and cluttered with handwriting, grading marks or stickers, and the authors call them Real-World Degraded Exam Images (RDEI). Converting such images into clean, question-level structured records is difficult. OCR discards structure such as question boundaries and reading order. MLLMs often invent text the pixels do not support and misplace question and figure regions.

MessToClean is a pipeline that works with any MLLM backbone and runs in three steps:
1. A two-stage fine-tuned RF-DETR detector locates question and figure boxes, which serve as pixel-aligned evidence.
2. The page hierarchy, reading order and question–figure links are rebuilt from these boxes.
3. A Generator-Verifier-Patcher loop compares MLLM output against the cropped regions and applies minimal corrections. It outputs Markdown together with JSONL logs of every edit.

The authors tested the pipeline with 12 backbones on RDEI-Exam, a new benchmark drawn from their own platforms. It improved stem fidelity, figure fidelity and refusal reliability in every case. This makes it practical to build auditable question banks from messy real-world photos.

## Key points
- **RDEI-Exam benchmark:** 12,472 real degraded exam images gathered across multiple regions and schools.
- **Gains across 12 MLLM backbones:**
  - stem consistency improved by 1.01–3.18%;
  - figure consistency improved by 0.50–49.16%;
  - refusal F1 improved by 1.06–10.88%.
- **Pipeline matters more than scale:** with MessToClean, the 8B Qwen3-VL outperformed its 30B sibling and roughly matched the 106B GLM-4.6v.
- **Detector training:** stage 1 uses synthetic occlusions with a "partial" class; stage 2 calibrates on real data.
  - Compared with single-stage training, the full recipe raised mAP_base@50:95 by 8.2 points and AP_Partial by 6.8 points, reaching 79.3 and 67.8.
  - Stage 2 alone added 29.7 points of AP_Partial.
- **Rotation sensitivity:** backbones averaged a StemSim of 65.7% at right-angle rotations but only 26.6% at oblique ones. The worst case was 11.3% at 135°.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [University of Science and Technology of China](https://md.research.elamur.ai/free/en/universities/university-of-science-and-technology-of-china/) — University, Hefei, China · 108 works · 112,698 citations
- [Nanyang Technological University](https://md.research.elamur.ai/free/en/universities/nanyang-technological-university/) — University, Singapore, Singapore · 156 works · 25,644 citations
- [Renmin University of China](https://md.research.elamur.ai/free/en/universities/renmin-university-of-china/) — University, Beijing, China · 114 works · 16,118 citations
- [Technical University of Munich](https://md.research.elamur.ai/free/en/universities/technical-university-of-munich/) — University, Munich, Germany · 55 works · 48,474 citations
- [Xidian University](https://md.research.elamur.ai/free/en/universities/xidian-university/) — University, Xi'an, China · 14 works · 1,357 citations

Also on the byline: [Kean University](https://ror.org/04wzzqn13).

📚 Shelves (2): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [👁️ Computer Vision and Pattern Recognition](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/) · 2,869 works

## Cite

Cite as: Jiayi Tuo, Cheng Tang, Zihan Wang, Chenyue Zhou, Yao Li, Yanbiao Ma, Chao Wang, Wei Dai, Mingxuan Wang, Shitong Qin and Ziwei Zhao. 2026. MessToClean: Evidence-Grounded Structure-Preserving Reconstruction for Real-World Degraded Exam Paper Images. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{tuo2026messtoclean,
  title = {{MessToClean: Evidence-Grounded Structure-Preserving Reconstruction for Real-World Degraded Exam Paper Images}},
  author = {Jiayi Tuo, Cheng Tang, Zihan Wang, Chenyue Zhou, Yao Li, Yanbiao Ma, Chao Wang, Wei Dai, Mingxuan Wang, Shitong Qin and Ziwei Zhao},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1871},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1871}
}
```

- [← previous MesoTree: Annotated Linguistic Resources for Quantitative Comparative Linguistic Analysis and NLP in Mesoamerica](https://md.research.elamur.ai/free/en/papers/pugh-2026-mesotree-annotated-linguistic-resources-quantitative/)
- [next → MetaBench: A Multi-task Benchmark for Assessing LLMs in Metabolomics](https://md.research.elamur.ai/free/en/papers/lu-2026-metabench-multi-task-benchmark-assessing/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/36/)

## References

50 references, 10 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Milan Aggarwal, Hiresh Gupta, Mausoom Sarkar, and Balaji Krishnamurthy. 2020. Form2Seq : A framework for higher-order form structure extraction. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 3830–3840, Online. Association for Computational L… [find](https://scholar.google.com/scholar?q=Milan+Aggarwal%2C+Hiresh+Gupta%2C+Mausoom+Sarkar%2C+and+Balaji+Krishnamurthy%2E+2020%2E+Form2Seq+%3A+A+framework+for+higher%2Dorder+form+structure+extraction%2E+In+Proceedings+of+the+2020+Conference+on+Empirical+Meth)
2. Utathya Aich, Shinjini Chakraborty, Deepan Sadhukhan, Swarnendu Ghosh, and Tulika Saha. 2026. Hilex: Image-based hierarchical layout extraction from question papers. In Document Analysis and Recognition – ICDAR 2025, pages 485–505, Cham. Springer Nature Switzerland. [find](https://scholar.google.com/scholar?q=Utathya+Aich%2C+Shinjini+Chakraborty%2C+Deepan+Sadhukhan%2C+Swarnendu+Ghosh%2C+and+Tulika+Saha%2E+2026%2E+Hilex%3A+Image%2Dbased+hierarchical+layout+extraction+from+question+papers%2E+In+Document+Analysis+and+Recogniti)
3. Daniel Bolya, Sean Foley, James Hays, and Judy Hoffman. 2020. Tide: A general toolbox for identifying object detection errors. In Computer Vision – ECCV 2020, pages 558–573, Cham. Springer International Publishing. [find](https://scholar.google.com/scholar?q=Daniel+Bolya%2C+Sean+Foley%2C+James+Hays%2C+and+Judy+Hoffman%2E+2020%2E+Tide%3A+A+general+toolbox+for+identifying+object+detection+errors%2E+In+Computer+Vision+%E2%80%93+ECCV+2020%2C+pages+558%E2%80%93573%2C+Cham%2E+Springer+Internation)
4. Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, and Sergey Zagoruyko. 2020. [End-to-end object detection with transformers](https://md.research.elamur.ai/free/en/papers/carion-2020-end-end-object-detection-transformers/). In Computer Vision – ECCV 2020, pages 213–229, Cham. Springer International Publishing.
5. Xi Chen, Xiao Wang, Soravit Changpinyo, AJ Piergiovanni, Piotr Padlewski, Daniel Salz, Sebastian Goodman, Adam Grycner, Basil Mustafa, Lucas Beyer, Alexander Kolesnikov, Joan Puigcerver, Nan Ding, Keran Rong, Hassan Akbari, Gaurav Mishra, Linting Xue, Ashish V Thapliyal, James Bradbury, and 10 other… — [PaLI: A Jointly-Scaled Multilingual Language-Image Model](https://md.research.elamur.ai/free/en/papers/chen-2022-pali-jointly-scaled-multilingual-language/)
6. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, Luke Marris, Sam Petulla, Colin Gaffney, Asaf Aharoni, Nathan Lintz, Tiago Cardal Pais, Henrik Jacobsson, Idan Szpektor, Nan-Jiang Jiang, and 3416 [other…](https://md.research.elamur.ai/free/en/papers/comanici-2025-gemini-2-5-pushing-frontier/)
7. Sagnik Das, Kunwar Yashraj Singh, Jon Wu, Erhan Bas, Vijay Mahadevan, Rahul Bhotika, and Dimitris Samaras. 2021. End-to-end piece-wise unwarping of document images. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 4268–4277. [find](https://scholar.google.com/scholar?q=Sagnik+Das%2C+Kunwar+Yashraj+Singh%2C+Jon+Wu%2C+Erhan+Bas%2C+Vijay+Mahadevan%2C+Rahul+Bhotika%2C+and+Dimitris+Samaras%2E+2021%2E+End%2Dto%2Dend+piece%2Dwise+unwarping+of+document+images%2E+In+Proceedings+of+the+IEEE%2FCVF+Inte)
8. Timo I. Denk and Christian Reisswig. 2019. {BERT}grid: Contextualized embedding for 2d document representation and understanding. In Workshop on Document Intelligence at NeurIPS 2019. [find](https://scholar.google.com/scholar?q=Timo+I%2E+Denk+and+Christian+Reisswig%2E+2019%2E+%7BBERT%7Dgrid%3A+Contextualized+embedding+for+2d+document+representation+and+understanding%2E+In+Workshop+on+Document+Intelligence+at+NeurIPS+2019%2E)
9. Tu Anh Dinh, Carlos Mullov, Leonard Bärmann, Zhaolin Li, Danni Liu, Simon Reiß, Jueun Lee, Nathan Lerzer, Jianfeng Gao, Fabian Peller-Konrad, Tobias Röddiger, Alexander Waibel, Tamim Asfour, Michael Beigl, Rainer Stiefelhagen, Carsten Dachs-bacher, Klemens Böhm, and Jan Niehues. 2024. SciEx: Benchma… [find](https://scholar.google.com/scholar?q=Tu+Anh+Dinh%2C+Carlos+Mullov%2C+Leonard+B%C3%A4rmann%2C+Zhaolin+Li%2C+Danni+Liu%2C+Simon+Rei%C3%9F%2C+Jueun+Lee%2C+Nathan+Lerzer%2C+Jianfeng+Gao%2C+Fabian+Peller%2DKonrad%2C+Tobias+R%C3%B6ddiger%2C+Alexander+Waibel%2C+Tamim+Asfour%2C+Michael+B)
10. Thao Do, Dinh Phu Tran, An Vo, and Daeyoung Kim. 2025. Reference-based post-ocr processing with llm for precise diacritic text in historical document recognition. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 27951–27959. [find](https://scholar.google.com/scholar?q=Thao+Do%2C+Dinh+Phu+Tran%2C+An+Vo%2C+and+Daeyoung+Kim%2E+2025%2E+Reference%2Dbased+post%2Docr+processing+with+llm+for+precise+diacritic+text+in+historical+document+recognition%2E+In+Proceedings+of+the+AAAI+Conference)
11. Zhentao He, Can Zhang, Ziheng Wu, Zhenghao Chen, Yufei Zhan, Yifan Li, Zhao Zhang, Xian Wang, and Minghui Qiu. 2025. Seeing is believing? mitigating OCR hallucinations in multimodal large language models. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. [find](https://scholar.google.com/scholar?q=Zhentao+He%2C+Can+Zhang%2C+Ziheng+Wu%2C+Zhenghao+Chen%2C+Yufei+Zhan%2C+Yifan+Li%2C+Zhao+Zhang%2C+Xian+Wang%2C+and+Minghui+Qiu%2E+2025%2E+Seeing+is+believing%3F+mitigating+OCR+hallucinations+in+multimodal+large+language+mod)
12. Thomas Hegghammer. 2022. Ocr with tesseract, amazon textract, and google document ai: a benchmarking experiment. Journal of Computational Social Science, 5(1):861–882. [find](https://scholar.google.com/scholar?q=Thomas+Hegghammer%2E+2022%2E+Ocr+with+tesseract%2C+amazon+textract%2C+and+google+document+ai%3A+a+benchmarking+experiment%2E+Journal+of+Computational+Social+Science%2C+5%281%29%3A861%E2%80%93882%2E)
13. Anwen Hu, Haiyang Xu, Liang Zhang, Jiabo Ye, Ming Yan, Ji Zhang, Qin Jin, Fei Huang, and Jingren Zhou. 2025. [mPLUG-DocOwl2: High-resolution compressing for OCR-free multi-page document understanding](https://md.research.elamur.ai/free/en/papers/hu-2025-mplug-docowl2-high-resolution-compressing/). In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1…
14. Shihua Huang, Zhichao Lu, Xiaodong Cun, Yongjun Yu, Xiao Zhou, and Xi Shen. 2025. Deim: Detr with improved matching for fast convergence. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 15162–15171. [find](https://scholar.google.com/scholar?q=Shihua+Huang%2C+Zhichao+Lu%2C+Xiaodong+Cun%2C+Yongjun+Yu%2C+Xiao+Zhou%2C+and+Xi+Shen%2E+2025%2E+Deim%3A+Detr+with+improved+matching+for+fast+convergence%2E+In+Proceedings+of+the+IEEE%2FCVF+Conference+on+Computer+Vision+a)
15. Ding Jia, Yuhui Yuan, Haodi He, Xiaopei Wu, Haojun Yu, Weihong Lin, Lei Sun, Chao Zhang, and Han Hu. 2023. Detrs with hybrid matching. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 19702–19712. [find](https://scholar.google.com/scholar?q=Ding+Jia%2C+Yuhui+Yuan%2C+Haodi+He%2C+Xiaopei+Wu%2C+Haojun+Yu%2C+Weihong+Lin%2C+Lei+Sun%2C+Chao+Zhang%2C+and+Han+Hu%2E+2023%2E+Detrs+with+hybrid+matching%2E+In+Proceedings+of+the+IEEE%2FCVF+Conference+on+Computer+Vision+and+)
16. Minghao Li, Tengchao Lv, Jingye Chen, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, and Furu Wei. 2023a. Trocr: Transformer-based optical character recognition with pre-trained models. In Proceedings of the AAAI conference on artificial intelligence, volume 37, pages 13094–13102. [find](https://scholar.google.com/scholar?q=Minghao+Li%2C+Tengchao+Lv%2C+Jingye+Chen%2C+Lei+Cui%2C+Yijuan+Lu%2C+Dinei+Florencio%2C+Cha+Zhang%2C+Zhoujun+Li%2C+and+Furu+Wei%2E+2023a%2E+Trocr%3A+Transformer%2Dbased+optical+character+recognition+with+pre%2Dtrained+models%2E+I)
17. Xiang Li, Wenhai Wang, Lijun Wu, Shuo Chen, Xiaolin Hu, Jun Li, Jinhui Tang, and Jian Yang. 2020. Generalized focal loss: Learning qualified and distributed bounding boxes for dense object detection. In Advances in Neural Information Processing Systems, volume 33, pages 21002–21012. Curran Associate… [find](https://scholar.google.com/scholar?q=Xiang+Li%2C+Wenhai+Wang%2C+Lijun+Wu%2C+Shuo+Chen%2C+Xiaolin+Hu%2C+Jun+Li%2C+Jinhui+Tang%2C+and+Jian+Yang%2E+2020%2E+Generalized+focal+loss%3A+Learning+qualified+and+distributed+bounding+boxes+for+dense+object+detection%2E+)
18. Yifan Li, Yifan Du, Kun Zhou, Jinpeng Wang, Xin Zhao, and Ji-Rong Wen. 2023b. [Evaluating object hallucination in large vision-language models](https://md.research.elamur.ai/free/en/papers/li-2023-evaluating-object-hallucination-large-vision/). In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 292–305, Singapore. Association for Computational Linguistic…
19. Zichao Li, Aizier Abulaiti, Yaojie Lu, Xuanang Chen, Jia Zheng, Hongyu Lin, Xianpei Han, Shanshan Jiang, Bin Dong, and Le Sun. 2025. READoc: A unified benchmark for realistic document structured extraction. In Findings of the Association for Computational Linguistics: ACL 2025, pages 21889–21905, Vi… [find](https://scholar.google.com/scholar?q=Zichao+Li%2C+Aizier+Abulaiti%2C+Yaojie+Lu%2C+Xuanang+Chen%2C+Jia+Zheng%2C+Hongyu+Lin%2C+Xianpei+Han%2C+Shanshan+Jiang%2C+Bin+Dong%2C+and+Le+Sun%2E+2025%2E+READoc%3A+A+unified+benchmark+for+realistic+document+structured+extra)
20. Minghui Liao, Zhaoyi Wan, Cong Yao, Kai Chen, and Xiang Bai. 2020. Real-time scene text detection with differentiable binarization. In Proceedings of the AAAI conference on artificial intelligence, volume 34, pages 11474–11481. [find](https://scholar.google.com/scholar?q=Minghui+Liao%2C+Zhaoyi+Wan%2C+Cong+Yao%2C+Kai+Chen%2C+and+Xiang+Bai%2E+2020%2E+Real%2Dtime+scene+text+detection+with+differentiable+binarization%2E+In+Proceedings+of+the+AAAI+conference+on+artificial+intelligence%2C+vo)
21. Wenhui Liao, Jiapeng Wang, Hongliang Li, Chengyu Wang, Jun Huang, and Lianwen Jin. 2025. Do-clayllm: An efficient multi-modal extension of large language models for text-rich document understanding. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 4038–4049. [find](https://scholar.google.com/scholar?q=Wenhui+Liao%2C+Jiapeng+Wang%2C+Hongliang+Li%2C+Chengyu+Wang%2C+Jun+Huang%2C+and+Lianwen+Jin%2E+2025%2E+Do%2Dclayllm%3A+An+efficient+multi%2Dmodal+extension+of+large+language+models+for+text%2Drich+document+understanding%2E+I)
22. Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dollar. 2017. [Focal loss for dense object detection](https://md.research.elamur.ai/free/en/papers/lin-2017-focal-loss-dense-object-detection/). In Proceedings of the IEEE International Conference on Computer Vision (ICCV).
23. Bodhisattwa Prasad Majumder, Navneet Potti, Sandeep Tata, James Bradley Wendt, Qi Zhao, and Marc Najork. 2020. Representation learning for information extraction from form-like documents. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6495–6504, Onl… [find](https://scholar.google.com/scholar?q=Bodhisattwa+Prasad+Majumder%2C+Navneet+Potti%2C+Sandeep+Tata%2C+James+Bradley+Wendt%2C+Qi+Zhao%2C+and+Marc+Najork%2E+2020%2E+Representation+learning+for+information+extraction+from+form%2Dlike+documents%2E+In+Proceedin)
24. Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, and Jingdong Wang. 2021. Conditional detr for fast training convergence. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 3651–3660. [find](https://scholar.google.com/scholar?q=Depu+Meng%2C+Xiaokang+Chen%2C+Zejia+Fan%2C+Gang+Zeng%2C+Houqiang+Li%2C+Yuhui+Yuan%2C+Lei+Sun%2C+and+Jingdong+Wang%2E+2021%2E+Conditional+detr+for+fast+training+convergence%2E+In+Proceedings+of+the+IEEE%2FCVF+International+)
25. OpenAI. 2024. Hello gpt-4o. OpenAI Blog. Accessed: 2026-01-05. [find](https://scholar.google.com/scholar?q=OpenAI%2E+2024%2E+Hello+gpt%2D4o%2E+OpenAI+Blog%2E+Accessed%3A+2026%2D01%2D05%2E)
26. OpenAI. 2025. Gpt-5 is here. OpenAI Product Page. Accessed: 2026-01-05. [find](https://scholar.google.com/scholar?q=OpenAI%2E+2025%2E+Gpt%2D5+is+here%2E+OpenAI+Product+Page%2E+Accessed%3A+2026%2D01%2D05%2E)
27. Pritika Ramu, Sijia Wang, Lalla Mouatadid, Joy Rim-chala, and Lifu Huang. 2024. re 2 : Region-aware relation extraction from visually rich documents. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Vol… [find](https://scholar.google.com/scholar?q=Pritika+Ramu%2C+Sijia+Wang%2C+Lalla+Mouatadid%2C+Joy+Rim%2Dchala%2C+and+Lifu+Huang%2E+2024%2E+re+2+%3A+Region%2Daware+relation+extraction+from+visually+rich+documents%2E+In+Proceedings+of+the+2024+Conference+of+the+North)
28. Joseph Redmon, Santosh Divvala, Ross Girshick, and Ali Farhadi. 2016. You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 779–788. [find](https://scholar.google.com/scholar?q=Joseph+Redmon%2C+Santosh+Divvala%2C+Ross+Girshick%2C+and+Ali+Farhadi%2E+2016%2E+You+only+look+once%3A+Unified%2C+real%2Dtime+object+detection%2E+In+Proceedings+of+the+IEEE+conference+on+computer+vision+and+pattern+reco)
29. Isaac Robinson, Peter Robicheaux, Matvei Popov, Deva Ramanan, and Neehar Peri. 2025. Rf-detr: Neural architecture search for real-time detection transformers. Preprint, arXiv:2511.09554. — [arXiv:2511.09554](https://arxiv.org/abs/2511.09554) [find](https://scholar.google.com/scholar?q=Isaac+Robinson%2C+Peter+Robicheaux%2C+Matvei+Popov%2C+Deva+Ramanan%2C+and+Neehar+Peri%2E+2025%2E+Rf%2Ddetr%3A+Neural+architecture+search+for+real%2Dtime+detection+transformers%2E+Preprint%2C+arXiv%3A2511%2E09554%2E+%E2%80%94+%5BarXiv%3A2511)
30. Katharine Sanderson. 2023. Gpt-4 is here: what scientists think. Nature, 615(7954):773. [find](https://scholar.google.com/scholar?q=Katharine+Sanderson%2E+2023%2E+Gpt%2D4+is+here%3A+what+scientists+think%2E+Nature%2C+615%287954%29%3A773%2E)
31. Peize Sun, Yi Jiang, Enze Xie, Wenqi Shao, Zehuan Yuan, Changhu Wang, and Ping Luo. 2021. What makes for end-to-end object detection? In Proceedings of the 38th International Conference on Machine Learning, volume 139 of Proceedings of Machine Learning Research, pages 9934–9944. PMLR. [find](https://scholar.google.com/scholar?q=Peize+Sun%2C+Yi+Jiang%2C+Enze+Xie%2C+Wenqi+Shao%2C+Zehuan+Yuan%2C+Changhu+Wang%2C+and+Ping+Luo%2E+2021%2E+What+makes+for+end%2Dto%2Dend+object+detection%3F+In+Proceedings+of+the+38th+International+Conference+on+Machine+Lea)
32. V Team, Wenyi Hong, Wenmeng Yu, Xiaotao Gu, Guo Wang, Guobing Gan, Haomiao Tang, Jiale Cheng, Ji Qi, Junhui Ji, Lihang Pan, Shuaiqi Duan, Weihan Wang, Yan Wang, Yean Cheng, Zehai He, Zhe Su, Zhen Yang, Ziyang Pan, and 74 others. 2026. [Glm-4.5v and glm-4.1v-thinking: Towards versatile multimodal reas…](https://md.research.elamur.ai/free/en/papers/team-2025-v-glm-4-5v-glm/)
33. Shengbang Tong, Zhuang Liu, Yuexiang Zhai, Yi Ma, Yann LeCun, and Saining Xie. 2024. [Eyes wide shut? exploring the visual shortcomings of multimodal llms](https://md.research.elamur.ai/free/en/papers/tong-2024-eyes-wide-shut-exploring-visual/). In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 9568–9578.
34. Dongsheng Wang, Natraj Raman, Mathieu Sibue, Zhiqiang Ma, Petr Babkin, Simerjot Kaur, Yulong Pei, Armineh Nourbakhsh, and Xiaomo Liu. 2024. DocLLM: A layout-aware generative language model for multimodal document understanding. In Proceedings of the 62nd Annual Meeting of the Association for Computa… [find](https://scholar.google.com/scholar?q=Dongsheng+Wang%2C+Natraj+Raman%2C+Mathieu+Sibue%2C+Zhiqiang+Ma%2C+Petr+Babkin%2C+Simerjot+Kaur%2C+Yulong+Pei%2C+Armineh+Nourbakhsh%2C+and+Xiaomo+Liu%2E+2024%2E+DocLLM%3A+A+layout%2Daware+generative+language+model+for+multimo)
35. Zifeng Wang, Zizhao Zhang, Jacob Devlin, Chen-Yu Lee, Guolong Su, Hao Zhang, Jennifer Dy, Vincent Perot, and Tomas Pfister. 2023. QueryForm: A simple zero-shot form entity query framework. In Findings of the Association for Computational Linguistics: ACL 2023, pages 4146–4159, Toronto, Canada. Assoc… [find](https://scholar.google.com/scholar?q=Zifeng+Wang%2C+Zizhao+Zhang%2C+Jacob+Devlin%2C+Chen%2DYu+Lee%2C+Guolong+Su%2C+Hao+Zhang%2C+Jennifer+Dy%2C+Vincent+Perot%2C+and+Tomas+Pfister%2E+2023%2E+QueryForm%3A+A+simple+zero%2Dshot+form+entity+query+framework%2E+In+Findings)
36. Zilong Wang, Yiheng Xu, Lei Cui, Jingbo Shang, and Furu Wei. 2021. LayoutReader: Pre-training of text and layout for reading order detection. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 4735–4744, Online and Punta Cana, Dominican Republic. Associa… [find](https://scholar.google.com/scholar?q=Zilong+Wang%2C+Yiheng+Xu%2C+Lei+Cui%2C+Jingbo+Shang%2C+and+Furu+Wei%2E+2021%2E+LayoutReader%3A+Pre%2Dtraining+of+text+and+layout+for+reading+order+detection%2E+In+Proceedings+of+the+2021+Conference+on+Empirical+Methods)
37. Zining Wang, Tongkun Guan, Pei Fu, Chen Duan, Qianyi Jiang, Zhentao Guo, Shan Guo, Junfeng Luo, Wei Shen, and Xiaokang Yang. 2025. Marten: Visual question answering with mask generation for multi-modal document understanding. In Proceedings of the Computer Vision and Pattern Recognition Conference,… [find](https://scholar.google.com/scholar?q=Zining+Wang%2C+Tongkun+Guan%2C+Pei+Fu%2C+Chen+Duan%2C+Qianyi+Jiang%2C+Zhentao+Guo%2C+Shan+Guo%2C+Junfeng+Luo%2C+Wei+Shen%2C+and+Xiaokang+Yang%2E+2025%2E+Marten%3A+Visual+question+answering+with+mask+generation+for+multi%2Dmoda)
38. Jiayang Wu, Wensheng Gan, Zefeng Chen, Shicheng Wan, and Philip S. Yu. 2023. Multimodal [large language models: A survey](https://md.research.elamur.ai/free/en/papers/minaee-2024-large-language-models-survey/). In 2023 IEEE International Conference on Big Data (BigData), pages 2247–2256.
39. Hangdi Xing, Feiyu Gao, Qi Zheng, Zhaoqing Zhu, Zirui Shao, and Ming Yan. 2025. Intelligent document parsing: Towards end-to-end document parsing via decoupled content parsing and layout grounding. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 19987–19998. [find](https://scholar.google.com/scholar?q=Hangdi+Xing%2C+Feiyu+Gao%2C+Qi+Zheng%2C+Zhaoqing+Zhu%2C+Zirui+Shao%2C+and+Ming+Yan%2E+2025%2E+Intelligent+document+parsing%3A+Towards+end%2Dto%2Dend+document+parsing+via+decoupled+content+parsing+and+layout+grounding%2E+In)
40. Xinhao Xu, Hui Chen, Mengyao Lyu, Sicheng Zhao, Yizhe Xiong, Zijia Lin, Jungong Han, and Guiguang Ding. 2025a. Mitigating hallucinations in multimodal large language models via image token attention-guided decoding. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the… [find](https://scholar.google.com/scholar?q=Xinhao+Xu%2C+Hui+Chen%2C+Mengyao+Lyu%2C+Sicheng+Zhao%2C+Yizhe+Xiong%2C+Zijia+Lin%2C+Jungong+Han%2C+and+Guiguang+Ding%2E+2025a%2E+Mitigating+hallucinations+in+multimodal+large+language+models+via+image+token+attention%2Dg)
41. Heng Yin, Yuqiang Ren, Ke Yan, Shouhong Ding, and Yongtao Hao. 2025. Rod-mllm: Towards more reliable object detection in multimodal large language models. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 14358– 14368. [find](https://scholar.google.com/scholar?q=Heng+Yin%2C+Yuqiang+Ren%2C+Ke+Yan%2C+Shouhong+Ding%2C+and+Yongtao+Hao%2E+2025%2E+Rod%2Dmllm%3A+Towards+more+reliable+object+detection+in+multimodal+large+language+models%2E+In+Proceedings+of+the+Computer+Vision+and+Pat)
42. Wenhao Yu, Tianrui Zong, He Zhang, Lin Yang, He Yang, Guanhao Wu, Ruohua Xu, Qinqin Yan, and Liangcai Gao. 2026. [find](https://scholar.google.com/scholar?q=Wenhao+Yu%2C+Tianrui+Zong%2C+He+Zhang%2C+Lin+Yang%2C+He+Yang%2C+Guanhao+Wu%2C+Ruohua+Xu%2C+Qinqin+Yan%2C+and+Liangcai+Gao%2E+2026%2E)
43. Icdar 2025 competition on understanding chinese college entrance exam papers. In Document Analysis and Recognition – ICDAR 2025, pages 523–536, Cham. Springer Nature Switzerland. [find](https://scholar.google.com/scholar?q=Icdar+2025+competition+on+understanding+chinese+college+entrance+exam+papers%2E+In+Document+Analysis+and+Recognition+%E2%80%93+ICDAR+2025%2C+pages+523%E2%80%93536%2C+Cham%2E+Springer+Nature+Switzerland%2E)
44. Chong Zhang, Yi Tu, Yixi Zhao, Chenshu Yuan, Huan Chen, Yue Zhang, Mingxu Chai, Ya Guo, Huijia Zhu, Qi Zhang, and Tao Gui. 2024a. Modeling layout reading order as ordering relations for visually-rich document understanding. In Proceedings of the 2024 Conference on Empirical Methods in Natural Langua… [find](https://scholar.google.com/scholar?q=Chong+Zhang%2C+Yi+Tu%2C+Yixi+Zhao%2C+Chenshu+Yuan%2C+Huan+Chen%2C+Yue+Zhang%2C+Mingxu+Chai%2C+Ya+Guo%2C+Huijia+Zhu%2C+Qi+Zhang%2C+and+Tao+Gui%2E+2024a%2E+Modeling+layout+reading+order+as+ordering+relations+for+visually%2Drich+)
45. Dacao Zhang, Kun Zhang, Le Wu, Mi Tian, Richang Hong, and Meng Wang. 2024b. Path-specific causal reasoning for fairness-aware cognitive diagnosis. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, KDD ’24, page 4143–4154, New York, NY, USA. Association for Comp… [find](https://scholar.google.com/scholar?q=Dacao+Zhang%2C+Kun+Zhang%2C+Le+Wu%2C+Mi+Tian%2C+Richang+Hong%2C+and+Meng+Wang%2E+2024b%2E+Path%2Dspecific+causal+reasoning+for+fairness%2Daware+cognitive+diagnosis%2E+In+Proceedings+of+the+30th+ACM+SIGKDD+Conference+on+K)
46. Jiaxin Zhang, Dezhi Peng, Chongyu Liu, Peirong Zhang, and Lianwen Jin. 2024c. Docres: A generalist model toward unifying document image restoration tasks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 15654– 15664. [find](https://scholar.google.com/scholar?q=Jiaxin+Zhang%2C+Dezhi+Peng%2C+Chongyu+Liu%2C+Peirong+Zhang%2C+and+Lianwen+Jin%2E+2024c%2E+Docres%3A+A+generalist+model+toward+unifying+document+image+restoration+tasks%2E+In+Proceedings+of+the+IEEE%2FCVF+Conference+on+)
47. Shifeng Zhang, Cheng Chi, Yongqiang Yao, Zhen Lei, and Stan Z. Li. 2020. Bridging the gap between anchor-based and anchor-free detection via adaptive training sample selection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). [find](https://scholar.google.com/scholar?q=Shifeng+Zhang%2C+Cheng+Chi%2C+Yongqiang+Yao%2C+Zhen+Lei%2C+and+Stan+Z%2E+Li%2E+2020%2E+Bridging+the+gap+between+anchor%2Dbased+and+anchor%2Dfree+detection+via+adaptive+training+sample+selection%2E+In+Proceedings+of+the+I)
48. Yian Zhao, Wenyu Lv, Shangliang Xu, Jinman Wei, Guanzhong Wang, Qingqing Dang, Yi Liu, and Jie Chen. 2024. Detrs beat yolos on real-time object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 16965–16974. [find](https://scholar.google.com/scholar?q=Yian+Zhao%2C+Wenyu+Lv%2C+Shangliang+Xu%2C+Jinman+Wei%2C+Guanzhong+Wang%2C+Qingqing+Dang%2C+Yi+Liu%2C+and+Jie+Chen%2E+2024%2E+Detrs+beat+yolos+on+real%2Dtime+object+detection%2E+In+Proceedings+of+the+IEEE%2FCVF+Conference+on+)
49. Yunqiu Xu, Linchao Zhu, and Yi Yang. 2025b. Mc-bench: A benchmark for multi-context visual grounding in the era of mllms. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 17675–17687. [find](https://scholar.google.com/scholar?q=Yunqiu+Xu%2C+Linchao+Zhu%2C+and+Yi+Yang%2E+2025b%2E+Mc%2Dbench%3A+A+benchmark+for+multi%2Dcontext+visual+grounding+in+the+era+of+mllms%2E+In+Proceedings+of+the+IEEE%2FCVF+International+Conference+on+Computer+Vision%2C+pa)
50. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, Chujie Zheng, Dayiheng Liu, Fan Zhou, Fei Huang, Feng Hu, Hao Ge, Haoran Wei, Huan Lin, Jialong Tang, and 41 others. 2025. [Qwen3 technical report](https://md.research.elamur.ai/free/en/papers/yang-2025-qwen3-technical-report/). Preprint, arXiv:2505.09388.

*Updated 1 November 2026*