Transkribus Community · PyLaia · Published March 16, 2023
The Dutchess I
Text Recognition★ Featured
Description
This public AI model for Transkribus is a powerful combination of four major handwriting recognition models, providing a comprehensive solution for deciphering historical documents from the Low Countries. It covers a wide range of handwritings from the 17th, 18th, and 19th centuries, utilizing a rich dataset contributed by the Nationaal Archief (National Archives of the Netherlands) and the Stadsarchief Amsterdam (Amsterdam City Archives), among other institutions.
The model is an amalgamation of the Amsterdam Notarial Super Model, Dutch_XVII_Century, IJsberg, and Dutch Mountains (18th Century), offering an extensive collection of notarial handwritings, chronicles, Dutch East India Company documents, and more.
Please avoid using this model as a basemodel when training custom models, or with the language model or SmartSearch option, as it will not work properly due to its size.
Try this model
Use this modelOpen in Transkribus
Very low error rate4.3% CER
Character Error Rate (CER) measures the percentage of characters incorrectly recognised. Lower is better. This model scored 4.3% on its validation set. As a rule of thumb, a CER below 10% is considered good for most handwritten material. This is a larger model trained on diverse material, which generally makes it more robust across different handwriting styles. That said, larger training sets also make it harder to push the CER down further.
Measured on the model's own validation data. Results on your documents may differ depending on handwriting style, document condition, language, and how closely your material resembles the training data.
Words11,693,499
Lines2,284,188
Training Pages49,205
Model ID50786