Datasets:

Lukaszl
/

eu-law-ocr-dataset-pl-100

Tasks:

Image-to-Text

Modalities:

Formats:

Languages:

Size:

Tags:

Libraries:

License:

Dataset card Data Studio Files Files and versions

xet

Community

Dataset Viewer

Auto-converted to Parquet Duplicate

Split (1)

train · 100 rows

Search is not available for this dataset

image image

EU Law OCR Dataset PL 100

A small public image-only OCR dataset containing 100 Polish-language pages of EU legal and regulatory documents.

This dataset is designed as a lightweight sample for:

OCR model evaluation
Polish OCR testing
document AI / VLM experiments

100 JPG page images
Polish-language documents
EU legal / regulatory content
image-only release (no text annotations)

What is NOT included

ground-truth transcriptions
OCR outputs
benchmark results
dataset splits

Purpose

This is not a full benchmark.

It is a compact dataset sample intended for:

quick OCR sanity checks
visual OCR comparison
testing OCR systems on Polish legal-style documents

Source

This dataset is a derived subset of:

https://huggingface.co/datasets/G4KMU/LEMUR

License / attribution

This dataset follows the licensing conditions of the original source:

https://creativecommons.org/licenses/by/4.0/

If you use this dataset, please also attribute the original LEMUR dataset.

Maintainer

clearOCR https://clearocr.com

Notes

small sample (100 images)
image-only dataset
intended for evaluation, not training

Downloads last month: 17