Skip to content
RAG Repo

YFCC100M (the Yahoo Flickr Creative Commons 100 Million dataset) gathers 100M photos and videos uploaded to Flickr, every one released by its owner under a Creative Commons licence. It is a long-standing benchmark for research in computer vision and multimedia retrieval.

Each item comes with rich metadata: tags, titles and descriptions, camera details, timestamps, and often geolocation. That context makes it useful well beyond image-text pairing, for example in place recognition or temporal analysis.

The catch is licensing. The Creative Commons licence differs by item, from the permissive CC BY through to NonCommercial variants, so the terms that apply to one photo may not apply to the next. Filter on the per-item licence before building anything you plan to ship.

image-textvideoflickrcreative-commonsmultimodalmetadata

Related sources