This repository contains Handwritten Text Recognition training data (layout segmentation and transcriptions) for the Sloane Lab HTR model. The HTR model is trained on the handwriting of Hans Sloane (1660-1753).
The Sloane Lab HTR model was developed in Transkribus (desktop client) and is based on the images and transcriptions of the Sloane’s Catalogue of Miscallanies (folio 3-152, recto and verso). The Catalogue of Miscallanies was chosen for the training data creation because it is known to be predominantly written by Sloane. The transcriptions for the training data were created by AEL Data Service as part of the Enlightenment Architectures project for the digital edition of the Catalogue of 'Miscellanies, Antiquities, Seals, Pictures, Mathematical Instruments, Agate handles, Agate Cups, Bottles, Spoons'
Sloane’s catalogues and their transcriptions contain historic racist language. We have written about the additional problems raised by digital technologies in Ortolja-Baird, A., & Nyhan, J. (2022) https://doi.org/10.1093/llc/fqab065. For more information, please see our website sloanelab.org.
Sloan, K., Ortolja-Baird, A., Nyhan, J., Pickering, V., & Fleming, M. (Eds.). (2019). Sir Hans Sloane’s Miscellanea which comprises his catalogues of Miscellanies, Antiquities, Seals, Pictures, Mathematical Instruments, Agate Handles and Agate Cups, Bottles, Spoons (Digital Edition). https://enlightenmentarchitectures.reconstructingsloane.org/cataloguemiscellanies/index.html
British Museum digital images of Miscellanea: © The Trustees of the British Museum. Shared under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) licence
For complete documentation please download: Git_Hub_Documentation.docx