README.md

metadata

license: mit
language:
  - ca
task_categories:
  - image-to-text
pretty_name: Esposalles-line
dataset_info:
  features:
    - name: image
      dtype: image
    - name: text
      dtype: string
  splits:
    - name: train
      num_examples: 2328
    - name: validation
      num_examples: 742
    - name: test
      num_examples: 757
  dataset_size: 3827
tags:
  - atr
  - htr
  - ocr
  - historical
  - handwritten

Esposalles - line level

Table of Contents

• Esposalles - line level
  • Table of Contents
  • Dataset Description
    • Dataset Summary
    • Languages
  • Dataset Structure
    • Data Instances
    • Data Fields

Dataset Description

• Homepage: The Esposalles Database
• Source: IEHHR
• Paper: The ESPOSALLES database: An ancient marriage license corpus for off-line handwriting recognition
• Point of Contact: TEKLIA

Dataset Summary

The Marriage Licenses ground-truth is compiled from the Marriage Licenses Books conserved at the Archives of the Cathedral of Barcelona.

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

Languages

All the documents in the dataset are written in Catalan.

Dataset Structure

Data Instances

{
  'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=1244x128 at 0x1A800E8E190,
  'text': 'donsella filla de Onofre Esquer morraler de Bara y'
}

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.