metadata

license: mit
language:
  - la
task_categories:
  - image-to-text
pretty_name: HOME-Alcar-line
dataset_info:
  features:
    - name: image
      dtype: image
    - name: text
      dtype: string
  splits:
    - name: train
      num_examples: 59969
    - name: validation
      num_examples: 7905
    - name: test
      num_examples: 6932
  dataset_size: 74806
tags:
  - atr
  - htr
  - ocr
  - historical
  - handwritten

HOME-Alcar - line level

HOME-Alcar - line level

Dataset Description

Homepage: HOME
Source: Arkindex
Point of Contact: TEKLIA

Dataset Summary

The HOME-Alcar (Aligned and Annotated Cartularies) dataset is a Medieval corpus. The 17 medieval manuscripts in this corpus are cartularies, i.e. books copying charters and legal acts, produced between the 12th and 14th centuries.

This dataset comes from the following publication: Stutzmann, D., Torres Aguilar, S., & Chaffenet, P. (2021). HOME-Alcar: Aligned and Annotated Cartularies [Data set]. Zenodo. https://doi.org/10.5281/zenodo.5600884

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

Languages

All the documents in the dataset are written in Latin.

Dataset Structure

Data Instances

{
  'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=4300x128 at 0x1A800E8E190,
  'text': 'quatre mille livres de tournoiz poiez, si com¬'
}

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.

Datasets:

Teklia
/

HOME-Alcar-line