achimrabus/crnn-ctc-prosta-mova

Prosta Mova HTR Model (Puigcerver CRNN)

A Handwritten Text Recognition (HTR) model for Prosta Mova — the Ruthenian language used in Ukrainian and Belarusian documents from the 16th to 18th centuries. Trained on early printed books from the Ostroh Printery and Academy; suitable for Ukrainian Church Slavonic texts as well as texts written in Prosta Mova. Based on the CNN + BiLSTM + CTC architecture introduced in Puigcerver (2017) and used as the backbone of PyLaia and Transkribus.

This is a clean-room PyTorch reimplementation of that published architecture (PyLaia-inspired). It does not use the PyLaia Python package and is not loadable by it — training and inference run via plain PyTorch (see Usage below).

Model Details

• Architecture: CNN encoder [12, 24, 48, 48 filters] + 3-layer Bidirectional LSTM (256 units) + CTC decoder (Puigcerver 2017)
• Input: Grayscale line images, normalized to 128 px height with aspect ratio preserved
• Output: UTF-8 text (East/Ukrainian Church Slavonic and Prosta Mova Cyrillic with diacritical marks)
• Vocabulary: 186 symbols (symbols.txt), including punctuation and combining diacritics
• Framework: Pure PyTorch — clean-room reimplementation of the Puigcerver (2017) architecture (PyLaia-inspired); the PyLaia package is not required

Performance

Metric	Value
Validation CER	3.77%
Training epochs	97
Training lines	58,843
Training pages	948
Validation lines	2,588
Validation pages	54

Training Data

Trained on images of early printed books transcribed and exported from Transkribus (see the corresponding Transkribus model page). The dataset covers Church Slavonic Ruthenian printings from the Ostroh Printery and Academy (end of the 16th to beginning of the 17th centuries).

Source texts:

• Ostroh Bible (1581)
• Kniga o postničestvě (1594)
• Margarit (1595)
• Otpis na list v boze velebnogo otca Ipatija volodimirs'kago i berestejskogo episkopa (1598)
• Apokrisis (1598–99)
• Pravilo istinnago života christianskogo (Psaltir z vozsliduvannjam) (1598)

The Transkribus training collection comprises 962 pages and 59,990 lines. Our CRNN-CTC model was trained on a corresponding export: 58,843 training lines (948 pages) and 2,588 validation lines (54 pages). The dataset was carefully preprocessed to correct EXIF rotation artifacts; aspect ratio preservation was applied to maintain character resolution.

The Transkribus model was created by Martin Meindl as part of the Continslav project, building on a generic East Church Slavonic printings model by Achim Rabus. Training data was prepared by Uliana Shtandenko and Alexandre Trébuchon. Model curated by Achim Rabus (Slavic Department, University of Freiburg).

Usage

Requirements

The inference code lives in polyscriptor and imports other modules from it, so run it from a clone of the repository. Install into a fresh virtual environment in one pip call; requirements-kraken.txt adds the Kraken segmentation used for full pages below (for a GPU install, see the polyscriptor README):

git clone https://github.com/achimrabus/polyscriptor
cd polyscriptor
python3 -m venv htr_env && source htr_env/bin/activate
pip install -r requirements.txt -r requirements-kraken.txt
hf download achimrabus/crnn-ctc-prosta-mova --local-dir models/crnn-ctc-prosta-mova

Inference

From the root of the clone:

from inference_pylaia_native import PyLaiaInference
from PIL import Image

# Load model
model = PyLaiaInference(
    checkpoint_path="models/crnn-ctc-prosta-mova/best_model.pt",
    syms_path="models/crnn-ctc-prosta-mova/symbols.txt"
)

# Transcribe a line image
image = Image.open("line_image.jpg")
text, confidence = model.transcribe(image)
print(text, f"(confidence {confidence:.2f})")

Note: Input should be a single text line image, not a full page. Preprocessing (grayscale conversion, height normalization, aspect ratio preservation) is handled automatically by inference_pylaia_native.py.

For full-page inference with automatic line segmentation, use batch_processing.py. The neural Kraken segmenter (kraken-blla) is strongly recommended; the default projection-based segmenter (hpp) is fast but can miss most lines on complex pages:

python batch_processing.py \
    --engine crnn-ctc \
    --model-path models/crnn-ctc-prosta-mova/best_model.pt \
    --segmentation-method kraken-blla \
    --input-folder images/ \
    --output-folder output/

Web Interface (recommended)

polyscriptor's main interface runs in the browser: upload page images or PDFs, segment them automatically (Kraken) or reuse existing PAGE XML, transcribe, correct lines inline and export TXT, CSV or PAGE XML. It runs on a laptop or on a remote server (via SSH tunnel), and CRNN-CTC models also work without a GPU.

The web interface picks up any folder under models/ that contains best_model.pt and symbols.txt, so after the download above it only needs to be started:

uvicorn web.polyscriptor_server:app --host 0.0.0.0 --port 8765
# open http://localhost:8765 and choose the model under CRNN-CTC

Desktop GUI

polyscriptor also ships PyQt6 desktop interfaces: transcription_gui_plugin.py for interactive single pages (automatic line segmentation, PAGE XML export) and polyscriptor_batch_gui.py for whole folders (uses existing PAGE XML files, e.g. from Transkribus, when available).

Intended Use

• Transcription of Prosta Mova and (Ukrainian) Church Slavonic early printed books
• Ukrainian and Belarusian historical document digitization (16th–18th centuries)
• Digital humanities research on early modern East Slavic texts

Limitations

• Optimized for Ostroh Printery-style printings; may underperform on other sources
• Full-page segmentation quality depends on the segmentation method used upstream

Citation

If you use this model in your research, please cite the architecture paper and this model:

@article{puigcerver2017multidimensional,
  title     = {Are Multidimensional Recurrent Layers Really Necessary for Handwritten Text Recognition?},
  author    = {Puigcerver, Joan},
  journal   = {Proceedings of the 14th IAPR International Conference on Document Analysis and Recognition (ICDAR)},
  year      = {2017},
  url       = {https://www.jpuigcerver.net/pubs/jpuigcerver_icdar2017.pdf}
}

@misc{rabus2026polyscriptor,
  title  = {Polyscriptor: Multi-Engine HTR Training \& Comparison Tool},
  author = {Rabus, Achim},
  year   = {2026},
  url    = {https://github.com/achimrabus/polyscriptor}
}