This Project repository contains models developed in project Potential of Artificial Intelligence for Digital Archives' Users that was co-funded by the European Union.

It is a cooperation work between National Archives of Finland (NAF), Central Archives for Finnish Business (ELKA) and South East University of Finland (Xamk).

The main model created in the project was a Finnish OCR model. The model is finetuned from PaddleOCR latin model, using a Finnish dataset from ELKA and combining that with Finnish synthetic data. All the data used in model training can be found from huggingface.