FIN-13K

This dataset is generated from https://github.com/sdrobac/ijdar-2020.

Dataset Summary

FIN-13K is a Finnish OCR dataset containing 13,037 image-text pairs extracted from historic Finnish newspapers and journals. The dataset is derived from the FIN-BERT dataset and is suitable for training and evaluating OCR models on Finnish text.

Supported Tasks

• Optical Character Recognition (OCR): Recognizing text from line-level images
• OCR Post-correction: For post-correction research, text and ocr_text are identical

Languages

Finnish (fi)

Dataset Structure

Data Instances

Each row contains:

• image: The textline image (PNG bytes)
• text: The ground truth transcription
• filename: The filename

Data Fields

Field	Type	Description
image	Image	Textline image
text	string	Ground truth transcription
filename	string	Filename

Data Splits

Split	Size
Train	13,037

Example

{
    "image": "images/00000.bin.png",
    "text": "takin yhdistystä kokoontukoot sunnuntaina",
    "filename": "00000"
}

Dataset Creation

Source Data

The dataset is derived from the FIN-BERT dataset, which contains historic Finnish newspapers and journals.

Personal and Sensitive Information

The dataset contains historic newspaper and journal content which may include historical references to individuals.

Licensing

CC-BY 4.0

Citation

@inproceedings{kiessling2019,
  title={An Open Annotated Dataset of Historic Finnish Newspapers and Journals for OCR and NLP},
  author={Kießling, Benjamin and Ebrahimi, Mohammad and Labbé, Mathieu and Oliveira, Sofia Ares},
  booktitle={ICDAR},
  year={2019}
}