Hello all,
Based on some recommendations on this list last year, we've been experimenting with using Transkribus to perform HTR of handwritten documents. I've noticed a high error rate in the transcriptions when using the pre-built models (Civil War era diaries), so we have been manually transcribing 40-60 pages and then training a model to recognize the rest of the text. However, even when using a base model, the character error rate of the new model is upwards of 25% – does anyone have advice for how to improve this number? I really like Transkribus as a program, and want to keep using it, but I'm wondering if others have had issues with the HTR, and if it's best for manual transcription.
Feel free to email me off list if you have golden advice to share! 🙂
Stay well,
----
Kayla Abner
(she/her)
Digital Scholarship Librarian
Digital Initiatives and Preservation
Library, Museums and Press
University of Delaware
[log in to unmask]<mailto:[log in to unmask]>
**The University of Delaware, a land grant institution, is located on land that was and continues to be vital to the web of life of the Nanticoke and Lenni-Lenape people. We express gratitude and honor the people who have inhabited, cultivated, and nourished this land for thousands of years, even after their attempted forced removal during the colonial era and early federal period. The University of Delaware also financially benefitted from the expropriation of Indigenous territories in the region colonially known as Montana. View the full Living Land Acknowledgement<https://sites.udel.edu/antiracism-initiative/committees/american-indian-and-indigenous-relations/living-land-acknowledgement/#Living_Land_Acknowledgement>.**
[cid:de387cd6-e4df-4e54-af50-b4ecd85272b6]
[cid:53d6dce9-a2e4-4aff-a953-7ba2313e1589]<[log in to unmask]&ep=bwmEmailSignature" target="_blank">https:[log in to unmask]&ep=bwmEmailSignature> Book time to meet with me<[log in to unmask]&ep=bwmEmailSignature" target="_blank">https:[log in to unmask]&ep=bwmEmailSignature>
|
