Skip to content

Dataset · jbarrow

Labeled form fields in document images

jbarrow/CommonForms

Large web dataset for detecting and classifying input fields on document scans and pages.

What's inside

Contains document page images with coordinate annotations and field types for filling: text inputs, selection buttons, and signatures. The dataset includes approximately 55,000 documents (metadata lists a range of 100k to 1M examples).

What you can do with it

  • Training object detectors for automated form and application parsing
  • Fine-tuning document analysis models and OCR pipelines
  • Creating auto-fill systems for PDF forms and web pages
Tasks
Object detection
Size
100K–1M rows
Downloads this month
2.8K
Likes
61

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("jbarrow/CommonForms")