Skip to content

Dataset · wanderkid

Image-based mathematical formula recognition

wanderkid/UniMER_Dataset

Over a million images of complex mathematical expressions for training OCR and image-to-text systems.

What's inside

A specialized collection for mathematical expression recognition (MER), consisting of the UniMER-1M training set (over 1 million examples) and a test set. It contains images of various formulas and their corresponding text representations. English and Chinese languages are supported.

What you can do with it

  • Training formula recognition models (OCR / image-to-text) for digitizing scientific papers and textbooks
  • Benchmarking multimodal models on complex mathematical expressions
Tasks
Image → text
Languages
English, Chinese
Size
1M–10M rows
Downloads this month
2.3K
Likes
30

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("wanderkid/UniMER_Dataset")