|
--- |
|
library_name: PyLaia |
|
license: mit |
|
tags: |
|
- PyLaia |
|
- PyTorch |
|
- atr |
|
- htr |
|
- ocr |
|
- historical |
|
- handwritten |
|
metrics: |
|
- CER |
|
- WER |
|
language: |
|
- 'no' |
|
datasets: |
|
- Teklia/NorHand_v3 |
|
pipeline_tag: image-to-text |
|
--- |
|
|
|
# PyLaia - NorHand v3 |
|
|
|
This model performs Handwritten Text Recognition in Norwegian. It was developed during the HUGIN-MUNIN project. |
|
|
|
## Model description |
|
|
|
The model has been trained using the PyLaia library on the [NorHand v3](https://zenodo.org/records/10255840) document images. |
|
|
|
Training images were resized with a fixed height of 128 pixels, keeping the original aspect ratio. |
|
|
|
| split | N lines | N horizontal lines | |
|
| ----- | ------: | -----------------: | |
|
| train | 224,173 | 223,971 | |
|
| val | 22,828 | 22,811 | |
|
| test | 1,573 | 1,573 | |
|
|
|
An external 6-gram character language model can be used to improve recognition. The language model is trained on the text from the NorHand v3 training set. |
|
|
|
## Evaluation results |
|
|
|
The model achieves the following results: |
|
|
|
| set | Language model | CER (%) | WER (%) | N lines | |
|
|:------|:---------------|:----------:|:-------:|----------:| |
|
| test | no | 7.52 | 22.99 | 1,573 | |
|
| test | yes | 6.36 | 18.11 | 1,573 | |
|
|
|
## How to use |
|
|
|
Please refer to the [documentation](https://atr.pages.teklia.com/pylaia/). |
|
|
|
## Cite us |
|
|
|
```bibtex |
|
@inproceedings{pylaia-lib, |
|
author = "Tarride, Solène and Schneider, Yoann and Generali, Marie and Boillet, Melodie and Abadie, Bastien and Kermorvant, Christopher", |
|
title = "Improving Automatic Text Recognition with Language Models in the PyLaia Open-Source Library", |
|
booktitle = "Submitted at ICDAR2024", |
|
year = "2024" |
|
} |
|
``` |
