HTRogène - Spain

Introduction

HTRogène is an exploratory project funded by Biblissima+, aiming to develop generic models for automatic transcription of medieval and early modern manuscripts.
This repository focuses on the Medieval Occitan corpus, providing ground-truth data for Handwritten Text Recognition (HTR) and layout segmentation.
The dataset is designed to support the creation of robust and reliable HTR models for Occitan manuscripts.

| Shelfmark | Links | Type | Century | Color Pages | Main Zones | Lines | Characters | Genre | |-----------------------------------------|----------------------------------------------------|--------|-----------|---------------|--------------|---------|--------------|------------| | London, British Library, Harley MS 3041 | B | prose | 14 | ✗ | 1 | 38 | 1634 | Narratives | | London, British Library, Harley MS 7403 | B | verse | 14 | ✗ | 10 | 175 | 3690 | Poetry | | Paris, BnF, fr. 13059 | B | verse | 14 | ✗ | 12 | 350 | 7414 | Narratives |

This dataset is part of CATMuS Medieval. More on CATMuS here.

Dataset Overview

The dataset comprises carefully selected manuscripts, each containing approximately 10 columns of text (equivalent to 5 bi-column pages or 10 single-column pages).
The data adheres to the Segmonto guidelines, ensuring consistency and compatibility with other datasets following the same standards.
Each image is accompanied by two XML files:

• Files suffixed with .chocomufin.xml are normalized for compliance with broader datasets.
• The other XML files contain repository-specific information.

We recommend using the normalized .chocomufin.xml files for most applications.

Total number of pages

65

Regions

• MainZone (101)
• DigitizationArtefactZone (28)
• NumberingZone (31)
• MarginTextZone (27)
• StampZone (1)
• RunningTitleZone (3)
• GraphicZone (2)
• DropCapitalZone (6)

Lines

• HeadingLine (68)
• DefaultLine (2808)
• InterlinearLine (4)

Funding and Support

This project is funded by Biblissima+, an observatory for medieval and Renaissance written cultural heritage.
Biblissima+ focuses on the study of the circulation of books and the transmission of texts from the 8th to 18th centuries.
Learn more at the Biblissima+ project page.

License

This dataset is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
You are free to share and adapt the material, provided appropriate credit is given.

Citation

If you use this dataset in your research, please cite it as follows:

Wiedner, M., & Couffignal, G., & Chagué, A., & Clérice, T. HTRogène, Occitan corpus of ground-truth for Handwritten Text Recognition and Layout Segmentation [Data set]. https://github.com/HTRogene/occitan

Acknowledgments

We extend our gratitude to the transcribers and supervisors who contributed to the creation of this dataset.

Special thanks to Biblissima+ for their financial support and commitment to advancing the study of medieval manuscripts.

For more information about the HTRogène project and other related resources, please visit the Biblissima+ project page.