| library_name: PyLaia | |
| license: mit | |
| tags: | |
| - PyLaia | |
| - PyTorch | |
| - Handwritten text recognition | |
| metrics: | |
| - CER | |
| - WER | |
| language: | |
| - en | |
| # English handwritten text recognition | |
| This model performs Handwritten Text Recognition in English. | |
| ## Model description | |
| The model has been trained using the PyLaia library on the [IAM](https://fki.tic.heia-fr.ch/databases/iam-handwriting-database) dataset. | |
| Training images were resized with a fixed height of 128 pixels, keeping the original aspect ratio. | |
| ## Evaluation results | |
| The model achieves the following results: | |
| | Split | CER (%) | WER (%) | Support | | |
| | ----- | ------- | ------- | ------- | | |
| | train | 0.32 | 1.26 | 6482 | | |
| | val | 6.50 | 19.12 | 1926 | | |
| | test | 7.68 | 19.82 | 1965 | | |
| A similar model was trained on the RWTH split, corresponding to the results published in [Key-value information extraction from full handwritten pages](https://arxiv.org/pdf/2304.13530.pdf). | |
| Results can be improved by combining PyLaia with a n-gram language model. | |
| ## How to use | |
| Please refer to the [documentation](https://atr.pages.teklia.com/pylaia/). | |