Search a semantic collection of 5.8 million artworks with The Last Museum's art discovery engine.
Semantic Art Search
github.com/KristianMSchmidt/semantic-art-search
Design Assets & Inspiration
Open-source projectOpen source
Search digital art by meaning with Semantic Art Search, an open-source engine for exploring artworks.
About Semantic Art Search
Semantic Art Search is an open-source search engine for exploring digitized art collections using natural-language queries. Rather than matching exact keywords in metadata, it ranks artworks by visual and semantic similarity, so results can surface works by theme, mood, style, or depicted content even when the title or description lacks the search term. It currently covers selected artworks from five institutions committed to open data: Statens Museum for Kunst, the Cleveland Museum of Art, Rijksmuseum Amsterdam, The Metropolitan Museum of Art, and the Art Institute of Chicago.
Searches can run across all museums or be limited to one or more, and results can be filtered by work type such as painting, print, or drawing, though filters still rely on metadata. Three matching modes are offered: Auto (visual plus title), Visual, and By title. Pasting an inventory number finds visually or thematically related works, and each result links back to the artwork's page at the source museum. The project runs on Jina CLIP v2 embeddings with vector nearest-neighbor search, and the repository includes Docker, Django, nginx and ETL components. Pricing is not stated.
Key features
- Natural-language search across digitized artworks by theme, mood or style
- Three modes: Auto (visual and title), Visual, and By title
- Paste an inventory number to find visually or thematically similar works
- Filter results by work type and by one or more museums
- Supports queries in multiple languages via Jina
- Uses Jina CLIP v2 embeddings with vector nearest-neighbor ranking
Best for: Developers, researchers and art enthusiasts who want an open-source, self-hostable way to search digitized museum collections by meaning rather than exact keywords.
Good for
- Open-source repository with Docker, Django and ETL components included
- Finds artworks by visual content even when metadata lacks the search term
- Links each result to its page at the source museum for full context
Watch out for
- Coverage is limited to selected artworks from five participating institutions
- Work-type and museum filters still depend on metadata rather than semantic matching
- Search relevance relies on a third-party embedding model (Jina CLIP v2)
Semantic Art Search at a glance
- Category
- Design & Creative › Design Assets & Inspiration
- Type
- Open-source project
- Pricing
- Not confirmed. Check the site
- Open source
- Yes · Source code
- Language
- English
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
Semantic Art Search alternatives
Tools and sites you could use instead, ranked by how closely they match.
Awesome Icons curates downloadable SVG, PNG, and font icon projects, grouped into general, logo, special, and framework sections.
Open source
24 alternatives
This repository curates design and UI resources, including stock photos, web templates, CSS frameworks, UI libraries, and tools.
Open source
24 alternatives
Create and edit React page layouts with Puck, a modular visual editor for React.
Open source
24 alternatives
Freesets curates free resources for web projects, including icons, images, illustrations, components, and libraries.
Open source
24 alternatives
Semantic Art Search finds art across museum collections using natural-language queries about subject, theme, or style.
Popular in Design Assets & Inspiration
Well-known listings from the same topic.
Arts and Cultureartsandculture.google.comFigmafigma.comUI Discoveryuidiscovery.comLS Graphicsls.graphicsToools.designtoools.designArtveeartvee.com
Semantic Art Search FAQ
How does this differ from a traditional keyword art search?
Traditional search returns works whose metadata contains the exact term. Semantic Art Search matches the intent of the query, so in Visual mode it can find a painting depicting a cat even if the word does not appear in the title or description.
Which museums are included?
Selected artworks from Statens Museum for Kunst, the Cleveland Museum of Art, Rijksmuseum Amsterdam, The Metropolitan Museum of Art, and the Art Institute of Chicago.
What do the search modes mean?
Auto is the default and combines visual and title search. Visual matches by visual appearance, and By title matches semantically against artwork titles.
Is something wrong or out of date?
Tell us and an editor will check it.
Own this website?
Claim this listing
Claim the listing and suggest updates yourself. An editor checks every claim first.