metadata

task_categories:
  - image-classification
  - image-segmentation
  - image-to-text
tags:
  - OCR
  - Text-Image Pairs
size_categories:
  - 10M<n<100M
license: osl-3.0
language:
  - en
pretty_name: Atlas PDF Image Cluster

Atlas PDF Image Cluster Dataset

Derives from the following Python Pipeline code: https://github.com/atlasunified/PDF-to-Image-Cluster

Dataset Description

This dataset is a collection of text extracted from PDF files, originating from various online resources. The dataset was generated using a series of Python scripts forming a robust pipeline that automated the tasks of downloading, converting, and managing the data.

Dataset Summary

Sample JPG

Corresponding JSON file with Bounding Box and Text data

Bounding box: [[0.10698689956331878, 0.008733624454148471], [0.7336244541484717, 0.008733624454148471], [0.7336244541484717, 0.06986899563318777], [0.10698689956331878, 0.06986899563318777]], Text: the Simchas Bais
Bounding box: [[0.013100436681222707, 0.12663755458515283], [0.7314410480349345, 0.12663755458515283], [0.7314410480349345, 0.1965065502183406], [0.013100436681222707, 0.1965065502183406]], Text: they are engaged in
Bounding box: [[0.0, 0.2445414847161572], [0.7379912663755459, 0.23580786026200873], [0.7379912663755459, 0.31222707423580787], [0.0, 0.31877729257641924]], Text: hey could become
Bounding box: [[0.008733624454148471, 0.36026200873362446], [0.7336244541484717, 0.36026200873362446], [0.7336244541484717, 0.425764192139738], [0.008733624454148471, 0.425764192139738]], Text: evil inclination still
Bounding box: [[0.004366812227074236, 0.48034934497816595], [0.31004366812227074, 0.4847161572052402], [0.31004366812227074, 0.5567685589519651], [0.004366812227074236, 0.5502183406113537]], Text: certainly
Bounding box: [[0.36899563318777295, 0.4890829694323144], [0.5480349344978166, 0.4890829694323144], [0.5480349344978166, 0.5524017467248908], [0.36899563318777295, 0.5524017467248908]], Text: men
Bounding box: [[0.5851528384279476, 0.4781659388646288], [0.740174672489083, 0.4781659388646288], [0.740174672489083, 0.5524017467248908], [0.5851528384279476, 0.5524017467248908]], Text: and
Bounding box: [[0.008733624454148471, 0.6004366812227074], [0.7336244541484717, 0.6004366812227074], [0.7336244541484717, 0.6681222707423581], [0.008733624454148471, 0.6681222707423581]], Text: e in separate areas.
Bounding box: [[0.9454148471615721, 0.6157205240174672], [0.9978165938864629, 0.6157205240174672], [0.9978165938864629, 0.6877729257641921], [0.9454148471615721, 0.6877729257641921]], Text: T
Bounding box: [[0.9519650655021834, 0.7532751091703057], [0.9978165938864629, 0.7532751091703057], [0.9978165938864629, 0.8078602620087336], [0.9519650655021834, 0.8078602620087336]], Text: 0
Bounding box: [[0.9475982532751092, 0.851528384279476], [0.9978165938864629, 0.851528384279476], [0.9978165938864629, 0.9235807860262009], [0.9475982532751092, 0.9235807860262009]], Text: fl\n

Sample JPG

Corresponding JSON file with Bounding Box and Text data

Bounding box: [[0.011570247933884297, 0.428099173553719], [0.9867768595041322, 0.428099173553719], [0.9867768595041322, 0.4677685950413223], [0.011570247933884297, 0.4677685950413223]], Text: tural person subiect to the reguirements laic
Bounding box: [[0.0049586776859504135, 0.5173553719008265], [0.9884297520661157, 0.5140495867768595], [0.9884297520661157, 0.5636363636363636], [0.0049586776859504135, 0.5669421487603306]], Text: priate, the provisions of sections 43 and 44;
Bounding box: [[0.009917355371900827, 0.6082644628099173], [0.9900826446280991, 0.6082644628099173], [0.9900826446280991, 0.6528925619834711], [0.009917355371900827, 0.6528925619834711]], Text: section 3. A person with no municipality of r
Bounding box: [[0.009917355371900827, 0.7041322314049587], [0.9917355371900827, 0.7041322314049587], [0.9917355371900827, 0.743801652892562], [0.009917355371900827, 0.743801652892562]], Text: ied by the authorities in their country of resi
Bounding box: [[0.0049586776859504135, 0.7917355371900826], [0.9917355371900827, 0.7950413223140496], [0.9917355371900827, 0.8396694214876033], [0.0049586776859504135, 0.8347107438016529]], Text:  firearm or firearm component in question ir

Sample JPG

Corresponding JSON file with Bounding Box and Text data

Bounding box: [[0.19349005424954793, 0.5334538878842676], [0.7902350813743219, 0.5370705244122965], [0.7902350813743219, 0.5822784810126582], [0.19349005424954793, 0.5786618444846293]], Text: Generic Drug Description
Bounding box: [[0.19529837251356238, 0.6274864376130199], [0.9909584086799277, 0.6274864376130199], [0.9909584086799277, 0.6708860759493671], [0.19529837251356238, 0.6708860759493671]], Text: Carboxymethylcellulose Sodium (

Supported Tasks and Use Cases

The primary use case of this dataset is to serve as training data for machine learning models that operate on text data. This may include, but is not limited to, text classification, information extraction, named entity recognition, and machine translation tasks.

Dataset Creation

This dataset was generated through a multi-stage Python pipeline designed to handle the downloading, conversion, and management of large datasets.

Primary URLs for downloading comes from ROM1504's dataset at the following link: http://3080.rom1504.fr/n/text/text38M/

Data Fields

As the dataset contains text extracted from PDF files from the common crawl. the data fields primarily include the extracted text and bounding box information.