How pabcablan/artemis works
Artemis is a visual and relational art-discovery platform. Its documented workflow collects Met Museum artwork records into MongoDB, enriches artwork context from Wikidata into Neo4j, and processes artwork images into Qdrant embeddings. In the Streamlit interface, a user uploads an image, gets visually similar works, and can inspect graph context and related artworks. The graph preserves these separate product stages and their stores; some cross-service data flows are shown from the README because their runtime wiring was not sampled.
Main entry points: enricher.py Artwork enricher · app.py Streamlit app · recommendations.py Visual recommendations · processor_usecase.py Image processor
Components 27
Artwork collection
- Artwork collector
metmuseum-feeder/application/usecases/collector.py - Met API adapter
metmuseum-feeder/infrastructure/adapters/met/adapter.py - Artwork repository
metmuseum-feeder/infrastructure/adapters/mongo/repository.py
Graph enrichment
- Artwork enricher
wikidata-feeder/application/usecases/enricher.py - Wikidata fetcher
wikidata-feeder/infrastructure/adapters/wikidata/fetcher.py - Artwork repository
wikidata-feeder/infrastructure/adapters/mongo/repository.py - Graph writer
wikidata-feeder/infrastructure/adapters/neo4j/client.py
Image processing
- Image processor
ia-image-processor/application/usecases/processor_usecase.py - Image repository
ia-image-processor/infrastructure/adapters/mongo/mongo_image_adapter.py - CLIP embeddings
ia-image-processor/infrastructure/adapters/clip/clip_service.py - Vector store
ia-image-processor/infrastructure/adapters/qdrant/qdrant_store.py
Art discovery UI
- Streamlit app
front/app.py - Image upload
front/infrastructure/adapters/streamlit/views/upload.py - Visual recommendations
front/application/usecases/recommendations.py - Image embedder
front/infrastructure/adapters/embedding.py - Vector search
front/infrastructure/adapters/vector_search.py - Search results
front/infrastructure/adapters/streamlit/views/results.py - Artwork details
front/application/usecases/details.py - Graph queries
front/infrastructure/adapters/graph/graph_query_adapter.py - Detail view
front/infrastructure/adapters/streamlit/views/detail.py
Data stores
- MongoDB
- Neo4j
- Qdrant
Other components
- User
- Met Museum API
- Wikidata
- Wikipedia
Connections 30
- Artwork collector → Met API adapter: uses provider
metmuseum-feeder/main.py - Artwork collector → Artwork repository: stores artworks
- Met API adapter → Met Museum API: fetches records
- Artwork repository → MongoDB: writes records
- Artwork enricher → Artwork repository: reads artworks
- Artwork enricher → Wikidata fetcher: fetches context
wikidata-feeder/main.py - Artwork enricher → Graph writer: writes graph
wikidata-feeder/main.py - Wikidata fetcher → Wikidata: queries
- Artwork enricher → Wikipedia: adds context
- Graph writer → Neo4j: writes graph
- Image processor → Image repository: loads images
ia-image-processor/main.py - Image processor → CLIP embeddings: creates embeddings
ia-image-processor/main.py - Image processor → Vector store: stores vectors
ia-image-processor/main.py - Image repository → MongoDB: reads artwork
- Vector store → Qdrant: stores vectors
- User → Streamlit app: uses interface
- Streamlit app → Image upload: renders upload
front/app.py - Streamlit app → Search results: renders results
front/app.py - Streamlit app → Detail view: renders detail
front/app.py - Image upload → Visual recommendations: submits image
- Visual recommendations → Image embedder: embeds image
front/application/usecases/recommendations.py - Visual recommendations → Vector search: searches vectors
front/application/usecases/recommendations.py - Visual recommendations → MongoDB: loads artwork
- Image embedder → CLIP embeddings: uses model
- Vector search → Qdrant: searches vectors
- Search results → User: shows matches
- Detail view → Artwork details: loads detail
- Artwork details → Graph queries: queries context
front/application/usecases/details.py - Artwork details → MongoDB: loads artwork
- Graph queries → Neo4j: queries relations