Dataset · common-canvas
High-resolution images (up to 4K) with descriptions
common-canvas/commoncatalog-cc-by
Massive dataset of images with Creative Commons licenses and synthetic captions.
What's inside
The collection includes between 10 and 100 million English-language images gathered in 2014 from Yahoo Flickr under Creative Commons licenses. The main feature is the high quality of originals up to 4K resolution. Each picture is accompanied by a generated (synthetic) text description.
What you can do with it
- Training and fine-tuning text-to-image generators (e.g., Stable Diffusion, Flux)
- Training models to generate detailed captions for images
- Tasks
- Text → image
- Languages
- English
- Size
- 10M–100M rows
- Downloads this month
- 67.2K
- Likes
- 45
How to load it
from datasets import load_dataset
ds = load_dataset("common-canvas/commoncatalog-cc-by")