Publications

A Comprehensive Comparison of Open-Source Libraries for Handwritten Text Recognition in Norwegian

Martin Maarand, Yngvil Beyer, Andre Kåsen, Knut T. Fosseide, Christopher Kermorvant

15th IAPR International Workshop, Das 2022, la Rochelle, France

In this paper, we introduce a database of historical handwritten documents in Norwegian, the first of its kind, allowing the development of handwritten text recognition models (HTR) in Norwegian.

Read our Paper

NorHand / Dataset for Handwritten Text Recognition in Norwegian

Yngvil Beyer, Andre Kåsen

Zenodo - May 20, 2022

The dataset comprises Norwegian letter and diary line images and text from 19th and early 20th century. It contains both the images, the manual transcriptions and the definitions of training and test sets for experiments.

Access the Dataset

Text line detection model for Doc-UFCN - Hugging Face

Mélodie Boillet, Christopher Kermorvant

Hugging Face - Nov 10, 2022

The Hugin-Munin line detection model predicts text lines from Hugin-Munin document images. It predicts two classes: vertical and horizontal text lines.

Use the Model

PyLaia v1.0.3 with language model support is released

Solène Tarride, Christopher Kermorvant

github.com - Dec 12, 2022

TEKLIA has released a new version of the HTR library PyLaia which support decoding with language models. TEKLIA is now the official maintainer of PyLaia.

Clone the repository

PyLaia HTR model publicly available on Hugging Face

Solène Tarride, Christopher Kermorvant

Hugging Face - Dec 18, 2022

TEKLIA has published on HuggingFace the PyLaia model for Norwegian handwriting recognition trained on the NorHand dataset

Use the Model

PyLaia v1.0.6 with temperature scaling support is released

Solène Tarride, Yoann Schneider, Christopher Kermorvant

github.com - Sep 12, 2023

TEKLIA has released a new version of the HTR library PyLaia which support temperature scaling to improve confidence scores.

Clone the repository