Dataset · UkrainianCatholicUniversity
Annotated handwritten Ukrainian texts
UkrainianCatholicUniversity/rukopys
The first major open dataset for Ukrainian handwritten text recognition spanning over a century.
What's inside
It contains document images with Ukrainian handwritten text—ranging from 1920s archival records to modern school notebooks. The dataset size ranges from 10k to 100k examples. The data is annotated for text region detection, type classification, and transcription.
What you can do with it
- Train OCR and HTR models for Ukrainian handwriting recognition
- Train text block detectors on historical and modern documents
- Build archive digitization systems for the Ukrainian language
- Tasks
- Object detection, Image → text
- Languages
- Ukrainian
- Size
- 10K–100K rows
- Downloads this month
- 5.8K
- Likes
- 28
Models trained on it
- Hukyl/trocr-kgb-archiveImage → text 188
- Hukyl/dfine-egret-x-rukopysObject detection 167
- Hukyl/rt-detrv2-heron-101-rukopysObject detection 156
- Hukyl/trocr-large-rukopys-hwImage → text 128
- ebinan92/Rukopys-OCR-4BImage text → text 115
- eldrar/cyrillic-large-handwritten-uk-rykopysImage → text 107
- Hukyl/trocr-large-rukopysImage → text 101
- Hukyl/trocr-rukopysImage → text 100
How to load it
from datasets import load_dataset
ds = load_dataset("UkrainianCatholicUniversity/rukopys")