Skip to content

Dataset · jordanparker6

Layout markup for scientific article pages

jordanparker6/publaynet

Page images from PubMed Central with layout block annotations for document structure recognition.

What's inside

The dataset contains between 10,000 and 100,000 page images of English-language research articles from the open PubMed Central collection. Layout markup for each document is specified using bounding boxes and polygonal segmentation. Annotations were automatically generated by comparing the PDF and XML versions of the articles.

What you can do with it

  • Training document layout analysis models
  • Improving OCR systems and scientific PDF parsers
Tasks
Image → text
Languages
English
Size
10K–100K rows
Downloads this month
4.4K
Likes
40

How to load it

from datasets import load_dataset

ds = load_dataset("jordanparker6/publaynet")