Harnessing Self-Supervised Features for Art Classification

Federico Melis, Davide Bilardello, Emanuele Prato, Evelyn Turri, Lorenzo Baraldi · 2026

Read Harnessing Self-Supervised Features for Art Classification by Federico Melis, Davide Bilardello, Emanuele Prato, Evelyn Turri, Lorenzo Baraldi (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-18974.txt.

Description

Federico Melis and four colleagues test whether self-supervised backbones beat supervised ones as feature extractors for artwork classification and retrieval, with paintings in focus. Classifying art is hard because fine-grained detail and abstract qualities jointly shape style and genre. Their evaluation across the DINO family and CLIP models, with several strategies and feature forms, finds consistent gains from self-supervised features. The work also maps how sorting and retrieval modules could serve real uses such as virtual-reality museum navigation.

Catalog record

CREATORFederico Melis, Davide Bilardello, Emanuele Prato, Evelyn Turri, Lorenzo Baraldi
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25333
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.18974

Browse the full library · About the Tower of Babel