Pith. sign in

REVIEW 3 major objections 5 minor 49 references

A Survey on Vietnamese Document Analysis and Recognition: Challenges and Future Directions

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This survey argues that Vietnamese OCR is held back less by model design than by diacritics, scarce data, and a lack of Vietnamese-specific LLMs.

desk verdict A useful but uneven survey of a niche field; the headline bottleneck claim about missing Vietnamese LLMs is internally contradicted by its own citations. read the letter →

arxiv 2506.05061 v1 pith:UN67UCCA submitted 2025-06-05 cs.CV

Anh Le , Thanh Lam , Dung Nguyen This is my paper
classification cs.CV
keywords VietnameseOCRdocumentanalysisandrecognitiondiacriticshandwrittentextlargelanguagemodelsdatasetscarcityscenepost-processing
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This survey maps the state of Vietnamese document analysis and recognition, from traditional feature-based OCR through deep learning and on to multimodal large language models. Its central claim is that the field's progress is gated by three interlocking bottlenecks: the linguistic difficulty of Vietnamese diacritics and word segmentation, a scarcity of large and diverse annotated datasets, and the absence of large language models trained specifically for Vietnamese documents. The paper argues that recognizing Vietnamese text reliably requires models that treat diacritical marks as first-class information, and that post-processing language models can substantially reduce the errors that remain after visual recognition. A sympathetic reader would take the survey as an evidence-organized argument that future gains will come from better datasets, Vietnamese-specific LLMs, and multimodal end-to-end document systems.

What carries the argument

The organizing mechanism is a task–method–dataset taxonomy: the survey separates recognition into printed, handwritten, and scene-text settings, catalogues the datasets and competitions that supply the only standardized evidence, and arrays methods along a pipeline of detection, recognition, and post-processing. Within that pipeline, the load-bearing component is the post-processing stage, where statistical language models, neural machine translation, and hill-climbing error correction reduce character error rate and word error rate on the VNOnDB benchmark; the survey uses these improvements to argue that linguistic modeling, not just visual recognition, carries the Vietnamese-specific performance gains.

What would settle it

Run the main cited recognition models on a single held-out Vietnamese test set and measure character error separately for diacritic-bearing versus plain letters; the claim that diacritics are a central bottleneck predicts a large, consistent gap favoring models with Vietnamese language post-processing, while a small or inconsistent gap would weaken it.

Watch

Extended reading notes

Core claim

The paper's core claim is that Vietnamese document analysis is not merely a special case of OCR but a distinct problem whose central difficulty is the diacritic system: identical letter sequences change meaning with tone marks, and the marks appear above and below characters, making them fragile under noise and complex backgrounds. It documents how competition benchmarks such as the ICFHR 2018 VOHTR challenge and the RIVF 2021 MC-OCR challenge produced the field's most comparable evidence, with best task results of character error rate 0.22 on mobile-captured receipts and sub-2% character error on online handwriting lines when strong language models were applied at post-processing. From this evidence the paper concludes that data scarcity and the lack of Vietnamese-specific LLMs, rather than raw model architecture, are the principal obstacles to higher accuracy and generalization.

Load-bearing premise

The survey treats the reported accuracy numbers from different papers as directly comparable even though each study used its own test data, metrics, and preprocessing, and that comparability is never checked.

Editorial extensions

If this is right

  • Standardizing datasets and evaluation protocols across Vietnamese OCR is necessary before method-to-method comparisons can be trusted.
  • Multimodal LLMs fine-tuned on Vietnamese data, such as the Vintern-1B model described in the survey, are a workable direction for OCR and document understanding despite data and compute constraints.
  • Post-processing with language models yields clear error reductions on handwriting, so OCR pipelines should treat linguistic correction as an integral stage rather than an optional extra.
  • Synthetic data generation and academic–industry collaboration are concrete paths to break the data-scarcity bottleneck.
  • A dedicated Vietnamese document AI system would need to combine OCR, layout analysis, and content extraction, pointing toward end-to-end multimodal architectures.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If a Vietnamese-specific LLM and a large unified benchmark existed, the relative value of visual recognition architectures versus language-based post-processing could be measured directly; the survey's structure implies that language-based correction would show the larger gain.
  • Diacritic-aware data augmentation, or a dedicated diacritic-error metric layered onto existing benchmarks, would make the paper's central linguistic difficulty measurable and testable.
  • The survey's dataset catalog suggests a testable extension: a model trained on synthetic Vietnamese documents and paired with a Vietnamese LLM corrector could match the best reported results, a claim the paper itself does not make.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper is a survey of Vietnamese document analysis and recognition (DAR). It reviews traditional, deep-learning, and LLM-based methods; describes two competitions (VOHTR 2018 and MC-OCR 2021); lists datasets; discusses OCR post-processing; and identifies challenges such as diacritics, word segmentation, data scarcity, model limitations, and community issues. It concludes that the main bottlenecks are data scarcity, poor model generalization, and the absence of Vietnamese-specific LLMs, and it proposes future research directions. The paper does not present new experiments; its contribution is a structured overview and a synthesis of reported results.

Significance. If taken as a reference map, the survey could be useful for researchers entering Vietnamese DAR, because it organizes a scattered literature across competitions, datasets, and post-processing methods, and it reports concrete benchmark numbers from VOHTR and MC-OCR. The paper is most valuable in its compilation of the two competition results and the dataset table, which are difficult to locate in one place. However, the survey's value as a synthesis is weakened by an internal contradiction on the LLM bottleneck and by the unstated coverage methodology. The central claim is defensible after reframing the bottleneck and adding a brief methodology note.

major comments (3)
  1. [§6, §3.3, §4.1] Section 6 (Conclusion) lists 'the absence of specialized large language models tailored for Vietnamese' as one of the key limitations, and Section 3.3 states that LLM application to Vietnamese document analysis 'is limited due to the scarcity of models specifically trained on Vietnamese data.' This is contradicted by the paper's own account in Section 4.1, where Vintern-1B is described as an efficient multimodal LLM 'optimized for tasks like OCR and document extraction in Vietnamese contexts,' fine-tuned on over 3 million image-question-answer pairs, and by the citation of PhoBERT in Section 2.3 as a Vietnamese-specific pretrained language model. The bottleneck should be reframed as a scarcity of large-scale annotated Vietnamese document images and the need for adaptation and fine-tuning, rather than the absence of Vietnamese-specific LLMs.
  2. [§1.2] The paper claims to provide a 'Comprehensive Review' but gives no search strategy, inclusion/exclusion criteria, database sources, or selection timeline. With 49 references and no audit trail, the comprehensiveness claim cannot be verified, and the coverage appears selective (for example, the survey does not discuss recent Vietnamese document-understanding models beyond Vintern-1B). Please add a methodology paragraph describing how works were identified and selected, or moderate the claim to 'a structured overview.'
  3. [Tables 1–4 and §2.2.2] The numeric results reported in Tables 1–4 and in the narrative of Sections 2.2–2.3 are presented without a general caveat about heterogeneous evaluation protocols. For instance, Table 4 compares post-processing methods on the same VNOnDB-OCR line-level dataset, which is valid, but the word accuracies on VinText/VnSceneText, the EM/F1 on ViOCR VQA, and the CER/WER on tetanus medical records in Section 2.2.2 come from different datasets and metrics and are not directly comparable. Add a sentence stating that cross-dataset numbers are indicative only and should not be used for ranking.
minor comments (5)
  1. [Throughout] There are numerous typos and formatting artifacts: 'T able 1' (Table 1 caption), 'F uture' (§1.2 bullet), 'W ord' and 'T onal' in §3.1 headings, 'T eam' in §2.2.1, 'Fine-T uning' and 'F ew-Shot' in §5.2, and 'This survey provide' in the abstract.
  2. [Table 3] The Cinnamon AI dataset row points to a GitHub repository by TomHuynhSG, not an official Cinnamon AI source; please verify the attribution.
  3. [§4.1 and §5.2] Some general claims lack citations, e.g., the statement that attention mechanisms in LLMs enable layout understanding, and the statement that self-supervised learning has been applied to improve few-shot performance.
  4. [§2.1] The rendering of diacritics such as 'Nˆ om' and 'm´ a'/'m` a' is broken in the PDF; please fix the font or encoding.
  5. [§2.3] Reference [46] is from 2012, but it is grouped with a 2024 LLM-based post-processing paper [45]; the sentence 'While the two OCR post-processing studies mentioned above [45, 46] evaluated their methods on own datasets' is potentially misleading about their contemporaneity and methodology.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the survey makes no quantitative derivation and its conclusions do not rest on any fitted parameter or self-citation chain.

full rationale

This paper is a literature survey and makes no quantitative predictions, fits no parameters, and derives no equations. Its central claims are descriptive: that Vietnamese DAR faces challenges in diacritics, word segmentation, data scarcity, model generalization, and LLM adaptation. These claims are supported by citations to the external literature, including benchmark competitions (VOHTR 2018, MC-OCR 2021), dataset papers, and method papers. The authors do cite their own prior work (e.g., Refs. [13], [14], [47]), but those self-citations are used only as entries in the survey of existing methods and datasets, not as load-bearing justifications for the survey's conclusions. Even the conclusion that 'the absence of specialized large language models tailored for Vietnamese' is a challenge is a qualitative observation contradicted in part by the paper's own discussion of Vintern-1B and PhoBERT; that is an internal-consistency or correctness issue, not circular reasoning. There is no step where a claimed output is equivalent to an input by definition, no fitted parameter renamed as a prediction, and no uniqueness theorem imported from the authors' prior work. The survey's value as a reference map may be limited by the lack of a formal search methodology and by comparing accuracy numbers across heterogeneous datasets, but these limitations affect reliability rather than circularity. Under the stated criteria, the honest finding is no significant circularity.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

This survey introduces no mathematical parameters, no new models, and no new data. Its burden is entirely descriptive, so the ledger contains only assumptions about the accuracy and representativeness of the material it reviews.

assumptions (3)
  • domain assumption The cited papers are accurately described and representative of the current state of Vietnamese DAR.
    The survey's value depends on the faithfulness of its secondary reporting; no independent verification is offered.
  • domain assumption Reported accuracy numbers across papers are comparable despite differing datasets and evaluation protocols.
    Tables 1-4 place results from VNOnDB, MC-OCR, and other benchmarks side by side as if they are directly comparable.
  • ad hoc to paper The taxonomy of traditional, deep learning, and LLM approaches is a useful organizing frame.
    The paper groups methods into these buckets (Sections 2, 4) and draws conclusions from that division.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Survey on Vietnamese Document Analysis and Recognition: Challenges and Future Directions." pith.science (2026). https://pith.science/paper/UN67UCCA

@misc{pith2026250605061,
  author       = {Pith},
  title        = {Pith review of: A Survey on Vietnamese Document Analysis and Recognition: Challenges and Future Directions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UN67UCCA}},
  note         = {Machine review of arXiv:2506.05061}
}
read the original abstract

Vietnamese document analysis and recognition (DAR) is a crucial field with applications in digitization, information retrieval, and automation. Despite advancements in OCR and NLP, Vietnamese text recognition faces unique challenges due to its complex diacritics, tonal variations, and lack of large-scale annotated datasets. Traditional OCR methods often struggle with real-world document variations, while deep learning approaches have shown promise but remain limited by data scarcity and generalization issues. Recently, large language models (LLMs) and vision-language models have demonstrated remarkable improvements in text recognition and document understanding, offering a new direction for Vietnamese DAR. However, challenges such as domain adaptation, multimodal learning, and computational efficiency persist. This survey provide a comprehensive review of existing techniques in Vietnamese document recognition, highlights key limitations, and explores how LLMs can revolutionize the field. We discuss future research directions, including dataset development, model optimization, and the integration of multimodal approaches for improved document intelligence. By addressing these gaps, we aim to foster advancements in Vietnamese DAR and encourage community-driven solutions.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

49 extracted references · 46 canonical work pages

  1. [1]

    International Journal on Document Analysis and Recognition (IJDAR) 24(1), 123–131 (2021)

    Van Hoai, D.P., Duong, H.-T., Hoang, V.T.: Text recognition for vietnamese identity card based on deep features network. International Journal on Document Analysis and Recognition (IJDAR) 24(1), 123–131 (2021)

    work page 2021

  2. [2]

    Pattern Recognition 78, 291–306 (2018)

    Nguyen, H.T., Nguyen, C.T., Bao, P.T., Nakagawa, M.: A database of uncon- strained vietnamese online handwriting and recognition experiments by recurrent neural networks. Pattern Recognition 78, 291–306 (2018)

    work page 2018

  3. [3]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

    Li, M., Lv, T., Chen, J., Cui, L., Lu, Y., Florencio, D., Zhang, C., Li, Z., Wei, F.: Trocr: Transformer-based optical character recognition with pre-trained models. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, pp. 13094–13102 (2023)

    work page 2023

  4. [4]

    In: European Conference on Computer Vision, pp

    Kim, G., Hong, T., Yim, M., Nam, J., Park, J., Yim, J., Hwang, W., Yun, S., Han, D., Park, S.: Ocr-free document understanding transformer. In: European Conference on Computer Vision, pp. 498–517 (2022). Springer

    work page 2022

  5. [5]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Nguyen, N., Nguyen, T., Tran, V., Tran, M.-T., Ngo, T.D., Nguyen, T.H., Hoai, M.: Dictionary-guided scene text recognition. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 7383–7392 (2021)

    work page 2021

  6. [6]

    Pattern recognition 28(9), 1399–1413 (1995) 16

    Bunke, H., Roth, M., Schukat-Talamazzini, E.G.: Off-line cursive handwriting recognition using hidden markov models. Pattern recognition 28(9), 1399–1413 (1995) 16

    work page 1995

  7. [7]

    In: 2010 Second International Conference on Computer Engineering and Applications, vol

    Nasien, D., Haron, H., Yuhaniz, S.S.: Support vector machine (svm) for english handwritten character recognition. In: 2010 Second International Conference on Computer Engineering and Applications, vol. 1, pp. 249–252 (2010)

    work page 2010

  8. [8]

    IJCSES 2(4), 258 (2008)

    Phuong, P.A., Tao, N.Q., Chi, L.: Speeding up isolated vietnamese handwritten recognition by combining svm and statistical features. IJCSES 2(4), 258 (2008)

    work page 2008

Show all 49 references
  1. [9]

    Dalat University Journal of Science 7(2), 142–152 (2017)

    Nga, P.T.T., Trang, N.T.H., Ph´ uc, N.V., Qu` y, T.D., B ` ınh, V.P.: Vietnamese text extraction from book covers. Dalat University Journal of Science 7(2), 142–152 (2017)

    2017

  2. [10]

    International Journal on Document Analysis and Recognition (IJDAR) 19, 49–64 (2016)

    Van Phan, T., Cong Nguyen, K., Nakagawa, M.: A nom historical document recog- nition system for digital archiving. International Journal on Document Analysis and Recognition (IJDAR) 19, 49–64 (2016)

    2016

  3. [11]

    In: 2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR), pp

    Nguyen, H.T., Nguyen, C.T., Nakagawa, M.: Icfhr 2018–competition on viet- namese online handwritten text recognition using hands-vnondb (vohtr2018). In: 2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR), pp. 494–499 (2018)

    2018

  4. [12]

    In: Proceedings of the Third Linguistic Annotation Workshop, pp

    Phuong-Thai, N., Xuan-Luong, V., Thi-Minh-Huyen, N., Van-Hiep, N., Hong- Phuong, L.: Building a large syntactically-annotated corpus of vietnamese. In: Proceedings of the Third Linguistic Annotation Workshop, pp. 182–185 (2009)

    2009

  5. [13]

    In: 2018 International Conference on Advanced Computing and Applications (ACOMP), pp

    Le, A.D., Nguyen, H.T., Nakagawa, M.: Recognizing unconstrained vietnamese handwriting by attention based encoder decoder model. In: 2018 International Conference on Advanced Computing and Applications (ACOMP), pp. 83–87 (2018)

    2018

  6. [14]

    SN Computer Science 1(1), 7 (2020)

    Le, A.D., Nguyen, H.T., Nakagawa, M.: An end-to-end recognition system for unconstrained vietnamese handwriting. SN Computer Science 1(1), 7 (2020)

    2020

  7. [15]

    In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp

    Vu, X.-S., Bui, Q.-A., Nguyen, N.-V., Nguyen, T.T.H., Vu, T.: Mc-ocr challenge: Mobile-captured image document recognition for vietnamese receipts. In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp. 1–6 (2021)

    2021

  8. [16]

    In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp

    Nguyen, C.M., Van Ngo, V., Nguyen, D.D.: Mc-ocr challenge 2021: Simple approach for receipt information extraction and quality evaluation. In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp. 1–4 (2021)

    2021

  9. [17]

    In: 2021 RIVF Inter- national Conference on Computing and Communication Technologies (RIVF), pp

    Nguyen, D.-C., Nguyen, T.-A., Nguyen, X.-C.: Mc-ocr challenge 2021: End-to-end system to extract key information from vietnamese receipts. In: 2021 RIVF Inter- national Conference on Computing and Communication Technologies (RIVF), pp. 1–5 (2021) 17

    2021

  10. [18]

    In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp

    Nguyen, H.V., Doan, L.B., Trinh, H.V., Phan, H.H., Thanh, T.M.: Mc-ocr chal- lenge 2021: Towards document understanding for unconstrained mobile-captured vietnamese receipts. In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp. 1–5 (2021)

    2021

  11. [19]

    In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp

    Bui, D.C., Truong, D., Vo, N.D., Nguyen, K.: Mc-ocr challenge 2021: Deep learn- ing approach for vietnamese receipts ocr. In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp. 1–6 (2021)

    2021

  12. [20]

    In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp

    Le, H., To, H., An, H., Ho, K., Nguyen, K., Nguyen, T., Do, T., Ngo, T.D., Le, D.-D.: Mc-ocr challenge 2021: An end-to-end recognition framework for viet- namese receipts. In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp. 1–6 (2021)

    2021

  13. [21]

    In: 2021 RIVF International Conference on Computing and Communication Technologies (RIVF), pp

    Hieu, T.B., Viet, H.D., Hiep, N.M., Anh, P.N.B., Bao, H.G., Anh, N.D., Bui, H.-P., Nguyen, T.H., Le Nguyen, P., Le, T.-L.: Mc-ocr challenge 2021: A multi- modal approach for mobile-captured vietnamese receipts recognition. In: 2021 RIVF International Conference on Computing an...

    2021

  14. [22]

    In: International Conference on Machine Learning, pp

    Tan, M., Le, Q.: Efficientnet: Rethinking model scaling for convolutional neu- ral networks. In: International Conference on Machine Learning, pp. 6105–6114 (2019)

    2019

  15. [23]

    arXiv preprint arXiv:1409.1556 (2014)

    Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014)

    2014 arXiv

  16. [24]

    https: //docs.ultralytics.com

    Ultralytics: YOLOv5: A state-of-the-art real-time object detection system. https: //docs.ultralytics.com. Accessed: 18 March 2025

    2025

  17. [25]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Baek, Y., Lee, B., Han, D., Yun, S., Lee, H.: Character region awareness for text detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 9365–9374 (2019)

    2019

  18. [26]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

    Wang, W., Xie, E., Song, X., Zang, Y., Wang, W., Lu, T., Yu, G., Shen, C.: Efficient and accurate arbitrary-shaped text detection with pixel aggregation net- work. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 8440–8449 (2019)

    2019

  19. [27]

    Advances in neural information processing systems 28 (2015)

    Ren, S., He, K., Girshick, R., Sun, J.: Faster r-cnn: Towards real-time object detec- tion with region proposal networks. Advances in neural information processing systems 28 (2015)

    2015

  20. [28]

    arXiv preprint arXiv:2009.09941 (2020) 18

    Du, Y., Li, C., Guo, R., Yin, X., Liu, W., Zhou, J., Bai, Y., Yu, Z., Yang, Y., Dang, Q., et al.: Pp-ocr: A practical ultra lightweight ocr system. arXiv preprint arXiv:2009.09941 (2020) 18

    2020 arXiv

  21. [29]

    In: Pro- ceedings of the IEEE/CVF International Conference on Computer Vision, pp

    Howard, A., Sandler, M., Chu, G., Chen, L.-C., Chen, B., Tan, M., Wang, W., Zhu, Y., Pang, R., Vasudevan, V., et al.: Searching for mobilenetv3. In: Pro- ceedings of the IEEE/CVF International Conference on Computer Vision, pp. 1314–1324 (2019)

    2019

  22. [30]

    https://github.com/pbcquoc/vietocr

    Pham, C.Q.B.: VietOCR. https://github.com/pbcquoc/vietocr. Accessed: 18 March 2025

    2025

  23. [31]

    In: 2020 25th International Conference on Pattern Recognition (ICPR), pp

    Yu, W., Lu, N., Qi, X., Gong, P., Xiao, R.: Pick: processing key information extraction from documents using improved graph learning-convolutional net- works. In: 2020 25th International Conference on Pattern Recognition (ICPR), pp. 4363–4370 (2021)

    2021

  24. [32]

    In: Findings of the Association for Computational Linguistics: EMNLP 2020, pp

    Nguyen, D.Q., Tuan Nguyen, A.: PhoBERT: Pre-trained language models for Vietnamese. In: Findings of the Association for Computational Linguistics: EMNLP 2020, pp. 1037–1042 (2020)

    2020

  25. [33]

    Journal of Military Science and Technology 90(90), 140–149 (2023)

    Huy, H.V., Thi Thanh Tan, N., Tao, N.Q.: Vietnamese text recognition in scene images using deep learning. Journal of Military Science and Technology 90(90), 140–149 (2023)

    2023

  26. [34]

    https://arxiv.org/abs/2404.18397

    Pham, H.Q., Nguyen, T.K.-B., Nguyen, Q.V., Tran, D.Q., Nguyen, N.H., Nguyen, K.V., Nguyen, N.L.-T.: ViOCR VQA: Novel Benchmark Dataset and Vision Reader for Visual Question Answering by Understanding Vietnamese Text in Images (2024). https://arxiv.org/abs/2404.18397

    2024 arXiv

  27. [35]

    ICIS 2023 Proceedings 2 (2023)

    Dinh, M.N., Le, M.T., Bui, T., Mai, M., Tran, L., Nguyen, N., Vo, T.H.: A vietnamese handwritten text recognition pipeline for tetanus medical records. ICIS 2023 Proceedings 2 (2023)

    2023

  28. [36]

    arXiv preprint arXiv:2408.12480 (2024)

    Doan, K.T., Huynh, B.G., Hoang, D.T., Pham, T.D., Pham, N.H., Nguyen, Q., Vo, B.Q., Hoang, S.N.: Vintern-1b: An efficient multimodal large language model for vietnamese. arXiv preprint arXiv:2408.12480 (2024)

    2024 arXiv

  29. [37]

    https://github.com/Nexdata-AI/ 4995-Vietnamese-OCR-Images-Data-Images-with-Annotation-and-Transcription

    Nexdata: 4,995 Vietnamese OCR Images Data - Images with Annotation and Transcription. https://github.com/Nexdata-AI/ 4995-Vietnamese-OCR-Images-Data-Images-with-Annotation-and-Transcription. Accessed: 27 March 2025

    2025

  30. [38]

    https://github.com/TomHuynhSG/Vietnamese-Handwriting-Recognition-OCR

    CinnamonAI: Vietnamese Handwriting Recognition Dataset by Cinnamon AI. https://github.com/TomHuynhSG/Vietnamese-Handwriting-Recognition-OCR. Accessed: 27 March 2025

    2025

  31. [39]

    https://www

    Kaggle: Vietnamese Handwritten OCR Dataset on Kaggle. https://www. kaggle.com/code/bomaich/vietnamese-handwritten-ocr-word-dataset. Accessed: 27 March 2025 19

    2025

  32. [40]

    https: //dataoceanai.com/datasets/ocr/vietnamese-ocr-image-dataset/

    DataoceanAI: Vietnamese OCR Image Corpus by DataoceanAI. https: //dataoceanai.com/datasets/ocr/vietnamese-ocr-image-dataset/. Accessed: 27 March 2025

    2025

  33. [41]

    In: 2022 RIVF International Conference on Computing and Communication Technologies (RIVF), pp

    Nguyen, N.H., Vo, D.T.D., Van Nguyen, K.: Uit-hwdb: Using transferring method to construct a novel benchmark for evaluating unconstrained handwriting image recognition in vietnamese. In: 2022 RIVF International Conference on Computing and Communication Technologies (RIVF), pp....

    2022

  34. [42]

    In: Proceedings of the 10th International Symposium on Information and Communication Technology, pp

    Nguyen, Q.-D., Le, D.-A., Zelinka, I.: Ocr error correction for unconstrained viet- namese handwritten text. In: Proceedings of the 10th International Symposium on Information and Communication Technology, pp. 132–138 (2019)

    2019

  35. [43]

    In: AIP Conference Proceedings, vol

    Nguyen, D., Le, A., Phan, M., Kromer, P., Zelinka, I.: Ocr error correction for viet- namese handwritten text using neural machine translation. In: AIP Conference Proceedings, vol. 2406 (2021)

    2021

  36. [44]

    Ieee Access 11, 58406– 58421 (2023)

    Nguyen, Q.-D., Phan, N.-M., Kr¨ omer, P., Le, D.-A.: An efficient unsupervised approach for ocr error correction of vietnamese ocr text. Ieee Access 11, 58406– 58421 (2023)

    2023

  37. [45]

    arXiv preprint arXiv:2410.13305 (2024)

    Do, T.: Reference-based post-ocr processing with llm for diacritic languages. arXiv preprint arXiv:2410.13305 (2024)

    2024 arXiv

  38. [46]

    In: Proceedings of the Workshop on Innovative Hybrid Approaches to the Processing of Textual Data, pp

    Hoang, C.D.V., Aw, A.: An unsupervised and data-driven approach for spell checking in vietnamese ocr-scanned texts. In: Proceedings of the Workshop on Innovative Hybrid Approaches to the Processing of Textual Data, pp. 36–44 (2012)

    2012

  39. [47]

    In: International Conference on Future Data and Security Engineering, pp

    Le, A., Mai, D.T.H., Lam, T.: A dataset of vietnamese documents for text detec- tion. In: International Conference on Future Data and Security Engineering, pp. 418–429 (2023)

    2023

  40. [48]

    https://github.com/docongminh/ VietNamese-OCR-DataGenerator

    Do, C.M.: Vietnamese OCR Data Generator. https://github.com/docongminh/ VietNamese-OCR-DataGenerator. Accessed: 18 April 2025

    2025

  41. [49]

    https://github.com/Belval/ TextRecognitionDataGenerator

    Belval, E.: Text Recognition Data Generator. https://github.com/Belval/ TextRecognitionDataGenerator. Accessed: 18 April 2025 20

    2025

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.