---
library_name: PyLaia
license: mit
tags:
- PyLaia
- PyTorch
- atr
- htr
- ocr
- modern
- handwritten
metrics:
- CER
language:
- zh
datasets:
- Teklia/CASIA
pipeline_tag: image-to-text
---

# PyLaia - CASIA-HWDB2

This model performs Handwritten Text Recognition in Chinese.

## Model description

The model was trained using the PyLaia library on the [CASIA-HWDB2](http://www.nlpr.ia.ac.cn/databases/handwriting/Offline_database.html) dataset.

Training images were resized with a fixed height of 128 pixels, keeping the original aspect ratio.

| set | lines | 
| :----- | ------: | 
| train | 33,425  |
| val   |  8,325  |
| test  | 10,449  |

An external 6-gram character language model can be used to improve recognition. The language model is trained on the text from the CASIA-HWDB2 training set.

## Evaluation results

The model achieves the following results:

| set   | Language model | CER (%)    | lines   |
|:------|:---------------| ----------:|----------:|
| test  | no             | 4.61       | 10,449    |
| test  | yes            | 1.53       | 10,449    |

## How to use?

Please refer to the [PyLaia documentation](https://atr.pages.teklia.com/pylaia/usage/prediction/) to use this model.

## Cite us!

```bibtex
@inproceedings{pylaia2024,
    author = {Tarride, Solène and Schneider, Yoann and Generali-Lince, Marie and Boillet, Mélodie and Abadie, Bastien and Kermorvant, Christopher},
    title = {{Improving Automatic Text Recognition with Language Models in the PyLaia Open-Source Library}},
    booktitle = {Document Analysis and Recognition - ICDAR 2024},
    year = {2024},
    publisher = {Springer Nature Switzerland},
    address = {Cham},
    pages = {387--404},
    isbn = {978-3-031-70549-6}
}
```