Skip to content

Dataset · wendlerc

12 million rendered handwriting texts on 3D paper

wendlerc/RenderedText

Synthetic dataset from Stability AI and LAION with English handwritten text images for OCR and generation.

What's inside

Inside are 12 million images sized 1024x1024 pixels with English handwritten text on a virtual 3D sheet of paper generated in Blender. The text features varying font size, color, and rotation, while the paper itself is rendered under random lighting.

What you can do with it

  • Training OCR systems to recognize handwriting at various angles and lighting conditions
  • Improving legible text rendering in image generation models
  • Pre-training multimodal encoders for working with documents
Tasks
Text → image, Image → text
Languages
English
Size
10M–100M rows
Downloads this month
20.3K
Likes
59

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("wendlerc/RenderedText")