Pith. sign in

REVIEW 3 major objections 5 minor 42 references

FP-THD: Full page transcription of historical documents

T0 review · 3 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read A two-stage pipeline transcribes full pages of 16th-century Latin print at 1.78% character error while keeping archaic symbols intact.

desk verdict Solid integration paper with a genuinely useful new dataset; the headline full-page numbers need a clear page-level holdout statement before they can be read as generalization. read the letter →

arxiv 2601.17040 v1 pith:UJHRRBEW submitted 2026-01-20 cs.CV cs.AI

H Neji , J Nogueras-Iso , J Lacasta , MÁ Latre , FJ García-Marco This is my paper · ORCID
classification cs.CVcs.AI
keywords historicaldocumenttranscriptionmaskedautoencodervisiontransformerlayoutanalysismedievalLatinOCRdiacriticpreservationfull-pagepipeline
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 paper attempts to show that full-page transcription of historical Latin documents can be done accurately without losing the archaic symbols that carry meaning. To do this it couples a layout-analysis model, which finds and crops text lines, with an OCR model built from a masked autoencoder and vision transformer trained on line images. On ten full pages of a 1585 printed law book the pipeline reports a character error rate of 1.78% and a word error rate of 4.50%, better than both an open OCR engine and a commercial transcription service, while preserving tildes, long-s forms, and ligatures. The same OCR component also reaches competitive error rates on two handwritten historical corpora without any post-processing or language-model correction.

What carries the argument

The load-bearing object is a two-component pipeline. The first component, ParseNet, is a convolutional encoder–decoder that detects baselines, ascenders, and descenders, outputs a PAGE-format XML layout (an XML standard for page geometry), and crops each line with geometric rectification to a standardized 50-pixel height. The second component is a masked autoencoder with vision transformer (MAE-ViT): a ResNet-18 feature extractor turns the line image into tokens, span masking (up to 8 consecutive tokens) replaces random masking as a regularizer, and CTC loss predicts the character sequence directly, without a language model. Because training data for Molino was annotated to preserve original

What would settle it

Take the ten full-page test images from the 1585 book, match them against the page list used to crop the 12,385 training lines, and retrain the OCR model on all non-overlapping pages; if the disjoint-holdout CER/WER rises materially above 1.78%/4.50%, the published advantage is partly memorization rather than generalization.

Watch

Extended reading notes

Core claim

The central claim is that the FP-THD pipeline—ParseNet for layout analysis followed by an MAE-ViT OCR model—can transcribe complete pages of medieval Latin printed material at character accuracy that outperforms existing tools. The authors created the Molino dataset from a 1585 edition of a civil-law commentary, trained the OCR on 12,385 line images, and tested on ten full pages. They report a CER of 1.78% and WER of 4.50%, with qualitative evidence that tildes and other diacritics are preserved better than by the Pero-OCR and ABBYY baselines.

Load-bearing premise

The full-page result depends on the ten test pages being disjoint from the pages used to build the 12,385 training line images, a separation the paper never explicitly states.

Editorial extensions

If this is right

  • If the reported numbers hold, the pipeline makes medieval Latin pages machine-searchable without modernizing their spelling, which matters for legal and historical scholarship.
  • The OCR component's performance on the Rodrigo and Bentham corpora suggests the same line recognizer can be reused across handwriting styles with only retraining, not a new architecture.
  • Because the system emits PAGE XML, Markdown, and plain text, the transcription can enter collaborative revision workflows with version control.
  • No post-processing or external language model is used, so the measured errors come from the visual model itself; future gains from decoding or context are additive.
  • At 4.5% word error rate the output is usable as input for the paper's planned downstream steps, such as automatic expansion of Latin abbreviations and translation.

Reading between the lines

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

  • Inference: The flagship Molino numbers should be read as provisional until the paper confirms that the ten full-page test images are not among the 12,385 training line images; Section 4.1, Table 2 does not state this split explicitly.
  • Inference: Because the Molino ground truth was generated from Pero-OCR output with manual correction (Appendix A), an independent evaluator could re-measure the pipeline against freshly hand-transcribed pages to rule out annotation bias toward any recognizer.
  • Inference: The same two-stage recipe—layout detection plus a span-masked ViT OCR trained on line images—could transfer to other early-modern scripts, such as German Fraktur, that also use abbreviation systems, provided enough annotated lines are created.
  • Inference: A direct extension would test whether preserved tildes and abbreviations actually improve downstream abbreviation-expansion and Latin-to-Spanish translation quality compared with modernized transcriptions.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 proposes FP-THD, a full-page transcription pipeline for historical documents, combining ParseNet-based layout analysis with a masked-autoencoder/Vision-Transformer OCR model. The OCR component is trained and evaluated on three datasets: two handwritten corpora (Rodrigo, Bentham) and a newly curated printed medieval-Latin corpus (Molino). The authors report line-level CER/WER on Rodrigo and Bentham, and a full-page comparison on 10 pages of the Molino 1585 book, claiming the best CER (0.0178) and WER (0.0450) over Pero-OCR and the BVPB/ABBYY transcription, with better preservation of tildes and archaic characters.

Significance. If the full-page Molino result is valid, the paper offers a practical, open, and effective pipeline for a niche but important task: faithful transcription of medieval Latin printed documents without post-processing or a language model. The strengths include a public code/data repository, no reliance on commercial OCR, competitive line-level results on Rodrigo, and a documented effort to preserve diacritics and special characters. However, the central full-page evaluation is currently under-specified: the page-level train/test split is not stated, and the Molino ground truth was produced with the same OCR engine family used as a comparison baseline. These issues are fixable with documentation, but they currently prevent the strongest claim from being accepted as evidence of generalization.

major comments (3)
  1. [§4.3, Table 6; Table 2 footnote] The full-page evaluation on 10 Molino test pages lacks a page-level holdout statement. §4.1 lists 37 pages from [36] among the Molino training sources, and §4.3 tests on 10 pages from the same [36] book. The paper never states that those 10 pages are disjoint from the pages used to generate the 12,385 training and 1,377 validation line images. If there is overlap, the reported CER=0.0178 and WER=0.0450 may reflect memorization rather than recognition. Please provide page identifiers, the selection procedure, and an explicit confirmation of disjointness.
  2. [Appendix A and Table 6] The Molino training/validation line images and ground-truth texts were generated using Pero-OCR, with manual correction of 'part of the transcriptions'. Pero-OCR is the same engine family as the comparison baseline in Table 6. The paper does not quantify which pages or characters were manually corrected, how many annotators were involved, or whether the 10 test-page ground truths were independently transcribed from the source images. Without this information, the comparison is partially circular: the model is trained on targets derived from the baseline and then evaluated against that same baseline. Please clarify the annotation workflow and, if possible, provide an independent test transcription.
  3. [§4.2, Tables 3 and 4] The line-level comparisons with Granell et al. [37], Chammas et al. [38], and HTR-JAND [39] are informative but not fully head-to-head unless the test splits are identical. The paper does not state whether the Rodrigo and Bentham test partitions match those used by the cited baselines. Please specify the split definitions and, where available, report results on the exact standard partitions so that 'competitive with state-of-the-art' can be verified.
minor comments (5)
  1. [§4.3, Table 6] 'ABBY' should be 'ABBYY' (the proprietary OCR engine). The same typo appears in the text of §4.3.
  2. [Section 1] There is a typo in 'symbols such assemployed' — likely intended as 'such as s employed' with a formatting issue around the long-s glyph.
  3. [Tables 1 and A.7] The configuration keys appear as 'LA YOUT_PARSER' and 'LA YOUT_PARSER_2'; this is likely a copy-paste/rendering issue but should be corrected for reproducibility.
  4. [Appendix A] The text refers to 'the input of our MSE model'; given the method section, this should probably be 'MAE model'.
  5. [§4.1] Rodrigo is described as 'Gothic typography' but this is a handwritten corpus; consider rewording to 'Gothic script' or 'Gothic handwriting' to avoid confusion.

Circularity Check

1 steps flagged · score 4.0 of 10

Molino ground truth is generated with the same Pero-OCR engine used as the comparison baseline, making the Table 6 superiority partially circular.

  1. self definitional [Appendix A; Table 6; §4.3]
    "In order not to do it completely manually, we took profit of the existing pre-trained models of ParseNet ... and Pero-OCR. ... Due to errors in the transcription of medieval Latin in Miguel del Molino’s book using Pero-OCR, we had to correct manually part of the transcriptions ... Our model achieves the best results, with the lowest CER (0.0178) and WER (0.0450), outperforming both Pero-OCR and ABBY transcriptions."

    The Molino ground-truth line images and texts used for training and for computing CER/WER were generated by Pero-OCR, the same engine listed as a comparison baseline in Table 6. The model is therefore trained to reproduce (corrected) Pero-OCR outputs, inheriting its character set, error patterns, and transcription conventions. Comparing FP-THD against Pero-OCR on pages from the same book is not an independent benchmark; the claimed 'outperforming' is partially self-referential. Manual correction reduces but does not remove the circularity, and its extent is not quantified.

full rationale

The only substantive circularity is in the Molino evaluation: the ground truth used both for training and for the Table 6 comparison was produced with the same Pero-OCR engine that serves as a baseline. This makes the central claim of outperforming Pero-OCR on Molino partially circular. However, the paper includes independent handwritten benchmarks (Rodrigo and Bentham) with standard test splits, and the Molino comparison also includes BVPB/ABBY, which partially mitigates the issue. I found no load-bearing self-citation chain: the authors' self-citations (e.g., [33]) concern a broader platform and are not used to justify the core recognition result. The Rodrigo and Bentham results are self-contained and reproducible from stated configurations. I also note, as a validation gap rather than a demonstrated circular step, that the paper never states whether the 10 full test pages from the Molino book are disjoint from the 37 pages used to build the training line images; if they overlap, the Table 6 numbers would reflect memorization, but the text does not allow me to prove such overlap. Overall, the Molino claim is weakened by the baseline-derived ground truth, but the independent benchmarks keep the paper from being wholly circular.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The pipeline itself introduces no new mathematical entities; the load-bearing assumptions are about data quality and evaluation independence, plus hand-chosen training constants. The most consequential is the Pero-OCR-derived ground truth and the unstated test/train split for Molino.

free parameters (5)
  • Line normalization height (LINE_HEIGHT) = 50 px
    Set in Table 1/Table A.7; every cropped line is rescaled to this height, a hand-chosen constant that affects OCR input.
  • Mask ratio = 0.4
    Table 5; fraction of input tokens masked during MAE-ViT training, chosen by hand.
  • Maximum span length = 8
    Section 3.3/Table 5; span-masking hyperparameter, chosen by hand.
  • Projection dimension = 8
    Table 5; ViT projection size, chosen by hand.
  • Training iterations = 100000
    Tables 3/5; training was stopped at a fixed iteration count for all datasets; no early-stopping or learning-rate schedule described.
assumptions (5)
  • standard math CTC loss with a learned character set is a valid sequence transcription objective for text-line images.
    Used by the OCR component (Section 3.3); standard for OCR but unproved in this paper.
  • domain assumption ParseNet layout is accurate enough that line-cropping errors do not dominate recognition errors.
    Full-page Molino test (Section 4.3) assumes layout is not the bottleneck; no propagation analysis.
  • domain assumption Molino ground truth derived from Pero-OCR with partial manual correction is an accurate training target.
    Appendix A; if false, comparisons in Table 6 are biased.
  • domain assumption The 10 Molino test pages are independent of the training pages and representative of the corpus.
    Table 2 footnote; holdout status not stated.
  • domain assumption The MAE-ViT architecture from Li et al. transfers to medieval Latin without architectural modification.
    Section 3.3 reuses [11] as-is; no novelty in OCR model.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FP-THD: Full page transcription of historical documents." pith.science (2026). https://pith.science/paper/UJHRRBEW

@misc{pith2026260117040,
  author       = {Pith},
  title        = {Pith review of: FP-THD: Full page transcription of historical documents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UJHRRBEW}},
  note         = {Machine review of arXiv:2601.17040}
}
read the original abstract

The transcription of historical documents written in Latin in XV and XVI centuries has special challenges as it must maintain the characters and special symbols that have distinct meanings to ensure that historical texts retain their original style and significance. This work proposes a pipeline for the transcription of historical documents preserving these special features. We propose to extend an existing text line recognition method with a layout analysis model. We analyze historical text images using a layout analysis model to extract text lines, which are then processed by an OCR model to generate a fully digitized page. We showed that our pipeline facilitates the processing of the page and produces an efficient result. We evaluated our approach on multiple datasets and demonstrate that the masked autoencoder effectively processes different types of text, including handwritten, printed and multi-language.

Figures

Figures reproduced from arXiv: 2601.17040 by the authors.

Figure 1
Figure 1. FP-THD architecture Overview: Layout Analysis and Masked Auto-encoder with Vision Trans [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗
Figure 2
Figure 2. Example text lines by datasets. The Rodrigo dataset is a corpus obtained from the digitisation of a handwritten book in old Castilian with chronicles of the history of Spain authored by an archbishop called Rodrigo in 1545 [30]. It contains 853 pages with Gothic typography, which are organised in 307 chapters [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 6 canonical work pages

  1. [36]

    M. d. Molino, Repertorium Fororum et Observantiarum Regni Aragonum: una pluribus cum determinationibus consilii iustitiae Aragonum practicis atquae 24 cautelis eisdem fideliter annexis, Dominici à Portonariis S. C. R. M. Regni Arag- oniae Typographi, Zaragoza, 1585. URLhttps://bvpb.mcu.es/es/consulta/registro.do?id=448556

  2. [37]

    Granell, E

    E. Granell, E. Chammas, L. Likforman-Sulem, C.-D. Martínez-Hinarejos, C. Mokbel, B.-I. Cîrstea, Transcription of spanish historical handwritten docu- ments with deep neural networks, Journal of Imaging 4 (1) (2018) 15

    2018

  3. [38]

    Chammas, C

    E. Chammas, C. Mokbel, Fine-tuning Handwriting Recognition systems with Temporal Dropout, arXiv preprint arXiv:2102.00511 (2021)

    arXiv 2021

  4. [39]

    Hamdan, A

    M. Hamdan, A. Rahiche, M. Cheriet, HTR-JAND: Handwritten Text Recogni- tion with Joint Attention Network and Knowledge Distillation, arXiv preprint arXiv:2412.18524 (2024)

    arXiv 2024

  5. [1]

    E. S. d. Santos Júnior, T. Paixão, A. B. Alvarez, Comparative performance of yolov8, yolov9, yolov10, and yolov11 for layout analysis of historical documents images, Applied Sciences 15 (6) (2025) 3164

    2025

  6. [2]

    Y . Zhou, S. Zuo, Z. Yang, J. He, J. Shi, R. Zhang, A review of document image enhancement based on document degradation problem, Applied Sciences 13 (13) (2023) 7855

    2023

  7. [3]

    H. Neji, J. Nogueras-Iso, J. Lacasta, M. B. Halima, A. M. Alimi, Adversarial au- toencoders for denoising digitized historical documents: the use case of incunab- ula, in: 2019 International Conference on Document Analysis and Recognition Workshops (ICDARW), V ol. 6, IEEE, 2019, pp. 31–34

    2019

  8. [4]

    H. Neji, M. Ben Halima, J. Nogueras-Iso, T. M. Hamdani, J. Lacasta, H. Chab- choub, A. M. Alimi, Doc-attentive-gan: Attentive gan for historical document denoising, Multimedia Tools and Applications 83 (18) (2024) 55509–55525

    2024

Show all 42 references
  1. [5]

    H. Neji, M. B. Halima, J. Nogueras-Iso, T. M. Hamdani, A. M. Qahtani, O. Almu- tiry, H. Dhahri, A. M. Alimi, Deep architecture for super-resolution and deblur- ring of text images, Multimedia Tools and Applications 83 (2) (2024) 3945–3961

    2024

  2. [6]

    Hamza, R

    S. Hamza, R. Menassel, C. Djeddi, Rseunet: residual squeeze and excitation u- net for restoration and binarization of historical documents, Multimedia Tools and Applications 83 (18) (2025) 1–24.doi:10.1007/s11042-025-21072-x. 20

    2025 doi

  3. [7]

    Ziran, M

    Z. Ziran, M. Mecella, S. Marinai, et al., Enhancing historical documents: Deep learning and image processing approaches, in: CEUR WORKSHOP PROCEED- INGS, V ol. 3937, CEUR-WS, 2025, pp. 0–0

    2025

  4. [8]

    Z. Yang, D. Peng, Y . Shi, Y . Zhang, C. Liu, L. Jin, Predicting the original appear- ance of damaged historical documents, in: Proceedings of the AAAI Conference on Artificial Intelligence, V ol. 39, 2025, pp. 9382–9390

    2025

  5. [9]

    Spina, Artificial intelligence in archival and historical scholarship work- flow: Hts and chatgpt, Umanistica Digitale (16) (2023).doi:10.6092/issn

    S. Spina, Artificial intelligence in archival and historical scholarship work- flow: Hts and chatgpt, Umanistica Digitale (16) (2023).doi:10.6092/issn. 2532-8816/17205. URLhttp://doi.org/10.6092/issn.2532-8816/17205

    2023 doi

  6. [10]

    S. T. Aguilar, Dual-Style Transcription of Historical Manuscripts based on Multi- modal Small Language Models with Switchable Adapters, in: International Con- ference on Applications of Natural Language to Information Systems, Springer, 2025, pp. 321–334.doi:10.1007/978-3-031-...

    2025 doi

  7. [11]

    Y . Li, D. Chen, T. Tang, X. Shen, Htr-vt: Handwritten text recognition with vision transformer, Pattern Recognition 158 (2025) 110967

    2025

  8. [12]

    Kodym, M

    O. Kodym, M. Hradiš, Page layout analysis system for unconstrained historic documents, in: Document Analysis and Recognition–ICDAR 2021: 16th Interna- tional Conference, Lausanne, Switzerland, September 5–10, 2021, Proceedings, Part II 16, Springer, 2021, pp. 492–506

    2021

  9. [13]

    Grüning, G

    T. Grüning, G. Leifert, T. Strauß, J. Michael, R. Labahn, A two-stage method for text line detection in historical documents, International Journal on Document Analysis and Recognition (IJDAR) 22 (3) (2019) 285–302

    2019

  10. [14]

    M. Diem, F. Kleber, S. Fiel, T. Grüning, B. Gatos, cBAD: ICDAR2017 com- petition on baseline detection, in: 2017 14th IAPR International Conference on 21 Document Analysis and Recognition (ICDAR), V ol. 1, IEEE, 2017, pp. 1355– 1360

    2017

  11. [15]

    Kohút, M

    J. Kohút, M. Hradiš, TS-Net: OCR trained to switch between text transcription styles, in: Document Analysis and Recognition–ICDAR 2021: 16th International Conference, Lausanne, Switzerland, September 5–10, 2021, Proceedings, Part IV 16, Springer, 2021, pp. 478–493

    2021

  12. [16]

    Rabaev, M

    I. Rabaev, M. Litvak, Recent advances in text line segmentation and base- line detection in historical document images: a systematic review, International Journal on Document Analysis and Recognition (IJDAR) 28 (2) (2025) 1–37. doi:10.1007/s10032-025-00526-w

    2025 doi

  13. [17]

    A. Lee, H. Yu, G. Min, An algorithm of line segmentation and reading order sort- ing based on adjacent character detection: a post-processing of ocr for digitization of chinese historical texts, Journal of Cultural Heritage 67 (2024) 80–91

    2024

  14. [18]

    W. Ma, H. Zhang, L. Jin, S. Wu, J. Wang, Y . Wang, Joint layout analysis, character detection and recognition for historical document digitization, in: 2020 17th In- ternational Conference on Frontiers in Handwriting Recognition (ICFHR), IEEE, 2020, pp. 31–36

    2020

  15. [19]

    K. He, X. Zhang, S. Ren, J. Sun, Deep residual learning for image recognition, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778

    2016

  16. [20]

    Van Koert, S

    R. Van Koert, S. Klut, T. Koornstra, M. Maas, L. Peters, Loghi: An end-to-end framework for making historical documents machine-readable, in: International Conference on Document Analysis and Recognition, Springer, 2024, pp. 73–88

    2024

  17. [21]

    N. A. Shiferaw, Z. L. Mayaluri, P. K. Sahoo, G. Panda, P. Jain, A. Rath, M. S. Islam, M. T. Islam, Handwritten amharic character recognition through transfer 22 learning: Integrating cnn models and machine learning classifiers, IEEE Access 13 (2025) 52134–52148.doi:10.1109/ACC...

    2025

  18. [22]

    K. He, X. Chen, S. Xie, Y . Li, P. Dollár, R. Girshick, Masked autoencoders are scalable vision learners, in: Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition, 2022, pp. 16000–16009

    2022

  19. [23]

    Campana, D

    S. Campana, D. Ferdani, H. Graf, G. Guidi, Z. Hegarty, S. Pescarin, F. Re- mondino, An open-source workflow for handwritten character recognition, in: Proceedings of Digital Heritage 2025, Eurographics/Digital Library, 2025, pp. 1–8, proceedings paper, identifier dh20253069.do...

    2025 doi

  20. [24]

    Pletschacher, Stefan and Antonacopoulos, Apostolos, The page (page analysis and ground-truth elements) format framework, in: 2010 20th International Con- ference on Pattern Recognition, IEEE, 2010, pp. 257–260

    2010

  21. [25]

    K. B. Kay, contributors, hOCR Specification, Version 1.2,https://kba. github.io/hocr-spec/, hTML-based open standard for representing OCR re- sults with layout information. (2018)

    2018

  22. [26]

    gov/standards/alto/, analyzed Layout and Text Object (ALTO) schema for OCR and layout metadata

    Library of Congress, ALTO XML Schema, Version 4.4,https://www.loc. gov/standards/alto/, analyzed Layout and Text Object (ALTO) schema for OCR and layout metadata. Maintained by the Library of Congress. (2022)

    2022

  23. [27]

    primaresearch.org/tools(2004)

    University of Salford Manchester, Tools — prima research lab,https://www. primaresearch.org/tools(2004)

    2004

  24. [28]

    URLhttps://www.tei-c.org/Vault/P5/4.9.0/doc/tei-p5-doc/en/ html/

    TEI Consortium, The TEI Guidelines P5, Version 4.9.0, TEI Consortium. URLhttps://www.tei-c.org/Vault/P5/4.9.0/doc/tei-p5-doc/en/ html/

  25. [29]

    Gruber, Markdown: A lightweight markup language (2004)

    J. Gruber, Markdown: A lightweight markup language (2004). URLhttps://daringfireball.net/projects/markdown/ 23

    2004

  26. [30]

    Granell, C.-D

    E. Granell, C.-D. Martínez-Hinarejos, The Rodrigo corpus (Nov. 2018).doi: 10.5281/zenodo.1490009. URLhttps://doi.org/10.5281/zenodo.1490009

    2018 doi

  27. [31]

    Causer, V

    T. Causer, V . Wallace, Building a volunteer community: results and findings from Transcribe Bentham, Digital Humanities Quarterly 6 (2) (2012) 1–14. URLhttps://www.digitalhumanities.org/dhq/vol/6/2/000125/ 000125.html

    2012

  28. [32]

    Bentham, Biography of Jeremy Bentham (1748–1832), english philosopher, ju- rist, and social reformer, regarded as the founder of modern utilitarianism (1832)

    J. Bentham, Biography of Jeremy Bentham (1748–1832), english philosopher, ju- rist, and social reformer, regarded as the founder of modern utilitarianism (1832). URLhttps://zenodo.org/records/44519

  29. [33]

    H. Neji, J. Nogueras-Iso, F. J. García-Marco, M. d. C. B. López, First steps towards a platform for the analysis of civil law documentary heritage, Scire: representación y organización del conocimiento 30 (1) (2024) 75–83.doi: 10.54886/scire.v30i1.5018

    2024 doi

  30. [34]

    Fueros, Zaragoza, 1667

    Herederos de Pedro Lanaja (Ed.), Fueros y Observancias del Reyno de Aragon de 1664-1667. Fueros, Zaragoza, 1667. URLhttps://derechoaragones.aragon.es/es/consulta/registro. do?id=600173

  31. [35]

    Delgado Echeverría, M

    J. Delgado Echeverría, M. Delgado Echeverría, Tomo III que acompaña a la edi- ción facsimilar con con estudio preliminar, traducciones, textos complementarios e índices., 1991. URLhttps://derechoaragones.aragon.es/bvda/es/consulta/ registro.do?id=600191

    1991

  32. [40]

    Cascianelli, V

    S. Cascianelli, V . Pippi, M. Maarand, M. Cornia, L. Baraldi, C. Kermorvant, R. Cucchiara, The LAM dataset: a novel benchmark for line-level handwritten text recognition, in: 2022 26th International Conference on Pattern Recognition (ICPR), IEEE, 2022, pp. 1506–1513

    2022

  33. [41]

    Marti, H

    U.-V . Marti, H. Bunke, The IAM-database: an English sentence database for offline handwriting recognition, International journal on document analysis and recognition 5 (1) (2002) 39–46

    2002

  34. [42]

    J. A. Sanchez, V . Romero, A. H. Toselli, E. Vidal, ICFHR2016 competition on handwritten text recognition on the READ dataset, in: 2016 15th International conference on frontiers in handwriting recognition (ICFHR), IEEE, 2016, pp. 630–635. Appendix A Curation of the Molino dat...

    2016

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.