Teklia/PELLET-Casimir-Marius-line

---
license: mit
language:
- fr
task_categories:
- image-to-text
pretty_name: PELLET Casimir Marius
dataset_info:
features:
- name: image
dtype: image
- name: text
dtype: string
splits:
- name: train
num_examples: 842
- name: validation
num_examples: 125
- name: test
num_examples: 122
dataset_size: 1089
tags:
- atr
- htr
- ocr
- historical
- handwritten
---

# PELLET Casimir Marius - Line level

## Table of Contents
- [PELLET Casimir Marius - Line level](#pellet-casimir-marius---line-level)
- [Table of Contents](#table-of-contents)
- [Dataset Description](#dataset-description)
- [Dataset Summary](#dataset-summary)
- [Languages](#languages)
- [Dataset Structure](#dataset-structure)
- [Data Instances](#data-instances)
- [Data Fields](#data-fields)
- [Usage with the PyLaia library](#usage-with-the-pylaia-library)

## Dataset Description

- **Homepage:** [Europeana](https://europeana.transcribathon.eu/documents/story/?story=121795/)
- **Point of Contact:** [TEKLIA](https://teklia.com)

## Dataset Summary

The PELLET Casimir Marius dataset includes 100 annotated French letters written between 1914 and 1918.
Annotations were done at line-level and all images do not have any text.

Note that all images are resized to a fixed height of 128 pixels.

### Languages

All the documents in the dataset are written in French.

## Dataset Structure

### Data Instances

```
{
'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=1684x128 at 0x1A800E8E190,
'text': 'LE HAVRE - panorama de la rue de Paris'
}
```

### Data Fields

- `image`: a PIL.Image.Image object containing the image. Note that when accessing the image column (using dataset[0]["image"]), the image file is automatically decoded. Decoding of a large number of image files might take a significant amount of time. Thus it is important to first query the sample index before the "image" column, i.e. dataset[0]["image"] should always be preferred over dataset["image"][0].
- `text`: the label transcription of the image.

## Usage with the PyLaia library

1. **Clone the repository** via
1. the Settings on the UI,
2. or `GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/datasets/Teklia/PELLET-Casimir-Marius-line`
2. The dataset is available in [PyLaia format](https://atr.pages.teklia.com/pylaia/usage/datasets/format/), in the `./pylaia` folder.

You can use this dataset to:
- train a new PyLaia model,
- assess your model's performance against this dataset.