ARTO KG: A Synthetic Artwork Dataset for Knowledge Enhanced Understanding Dataset Description ARTO KG is a large scale synthetic artwork dataset that bridges visual content and structured knowledge through ontology guided automated generation. Each artwork is annotated with comprehensive RDF knowledge graphs aligned with the ARTO ontology. Dataset Summary Total Artworks : 10,108 high resolution images (1024×1024) Object Instances : 39,878 (average 3.95 per artwork) RDF Triples : 1,056,970 (average 104.6 per artwork) Spatial Relations : 33,579 (90.6% of all relations) Semantic Interactions : 3,474 (9.4% of all relations) Artistic Styles : 5 (Baroque, Neoclassicism, Impressionism, Post Impressionism, Chinese Ink Painting) Total Size : ~14GB Supported Tasks Scene Graph Generation : Rich spatial and semantic relationships between objects Visual Question Answering : SPARQL queryable knowledge graphs Object Detection : Bounding box annotations for all objects Style Classification : Multi style artwork classification Semantic Retrieval : Knowledge graph based artwork retrieval Compositional Understanding : Complex object interactions and arrangements Dataset Structure Data Instances Each…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy