Search Papers | Poster Sessions | All Posters
Poster C81 in Poster Session C - Friday, August 9, 2024, 11:15 am – 1:15 pm, Johnson Ice Rink
Exploring the similarity space of visual art
Edward Vessel1 (), Andrew Frankel2, Sunil Ale1, Colin Conwell3; 1City College, The City University of New York, 2Graduate Center, The City University of New York, 3The Johns Hopkins University
What is the similarity structure of visual art? Using DreamSim, a variant on a CLIP neural network model optimized to produce human-like similarity scaling, we investigated the distinctiveness of artworks considering both art movements and artists, two dominant forms of category information organizing visual art. Art movements differed in their distinctiveness from other movements and also in the degree to which different artists within each movement were separable. This work highlights the promise of using linguistically-informed similarity spaces for understanding the impact of the arts on cognition.
Keywords: computational aesthetics visual artwork similarity