SMK Image-Text
An image–text dataset built from digitized SMK (Statens Museum for Kunst) artworks, with Danish and English descriptions.
vision-languagedatasetsdanish
An image–text dataset built from the digitized collection of Statens Museum for Kunst (SMK), Denmark’s national gallery.
Each artwork is paired with curatorial descriptions in Danish and, where available, English — giving downstream vision–language models a bilingual signal grounded in real museum metadata rather than generic web alt-text.
Why bother
Most Danish vision–language data is machine-translated from English webscrapes. That works for some things, but it’s not great if you want models that can talk about Danish-specific visual culture (paintings, sculpture, period furniture, etc.). SMK’s catalog is curated, well-described, and openly accessible via their API — so it makes a clean source of culturally-grounded Danish image–text pairs.
What’s in the dataset
- One row per artwork (paintings, drawings, sculpture, etc.) with the image URL, Danish description, and English description when SMK has one.
- Provenance metadata: artwork ID, artist, period, classification, license.
- Filtered to images that are public-domain or otherwise safely redistributable.