dsilvera · PyLaia · Published April 10, 2025

SPJCL 17C 5.1

Text Recognition

Description

Model created to transcribe Bevis Marks archive account books hand written in Portuguese and Spanish with some Romanized Hebrew for years 5436-5438, 5439-5440, 5440-5441, 5441-5442, 5445, 5453, 5454. (1676-1694) Model built without base model and using Ground Truth only.

Try this model

SPJCL 17C 5.1
Use this modelOpen in Transkribus
Very low error rate4.23% CER

Character Error Rate (CER) measures the percentage of characters incorrectly recognised. Lower is better. This model scored 4.23% 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.

Words188,463
Lines82,847
Training Pages819
Model ID320833
Languages
Portuguese