Hint for the OCR - recognized text has a specific font

[Dmitry Yatcenko]

I'm developing a program for translating card games. It frequently uses Gothic and grotesque fonts, which are often recognized with errors.
The project includes otf/ttf files for these fonts.
Is it possible to dynamically indicate to the OCR that the recognized text has a specific font (or font directory) without training the model?

[Zdenko Podobny]

Hi,

no this is not possible. If you have a very specific font, you need to finetune the model.

Zdenko

st 15. 4. 2026 o 11:41 Dmitry Yatcenko <yatc...@gmail.com> napísal(a):

I'm developing a program for translating card games. It frequently uses Gothic and grotesque fonts, which are often recognized with errors.
The project includes otf/ttf files for these fonts.
Is it possible to dynamically indicate to the OCR that the recognized text has a specific font (or font directory) without training the model?

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-oc...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/tesseract-ocr/0fdc4db1-9e3f-42e8-8af3-30a58fc2f2f9n%40googlegroups.com.