Skip to content

Dataset · mathieu1256

FATURA2 invoice dataset for NER

mathieu1256/FATURA2-invoices

Dataset of 20,000 invoice images with named entity recognition markup and JSON annotations.

What's inside

Includes 10,000 white-background and 10,000 colored-background images generated from 50 templates, along with corresponding JSON annotation files. Contains fields with images, NER tags, words, and bounding boxes (bboxes) in English.

What you can do with it

  • Training text recognition and OCR models on documents
  • Information extraction and data structuring from invoices
  • Training multimodal LLMs for document analysis tasks
Tasks
Embeddings
Languages
English
Size
10K–100K rows
Downloads this month
380
Likes
22

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("mathieu1256/FATURA2-invoices")