| KEYWORD |
Deep Style: Exploiting Neural Networks to understand the element of styles in paintings (𝓐𝓡𝓣ificial Intelligence)
Riferimenti GIOVANNI SQUILLERO
Riferimenti esterni Alberto Tonda
Pietro Barbiero
Gruppi di ricerca DAUIN - GR-05 - ELECTRONIC CAD & RELIABILITY GROUP - CAD
Descrizione Understanding whether two paintings are “similar” or simply “related” currently requires specific knowledge, expertise, and years of study; in the future, it may require a few seconds and a neural network.
The thesis project aims at exploiting the latent space of a deep neural network (DNN) together with the brand new idea of “logic layers'' developed for “Explainable AI” (XAI). A modified DNN will be used to discover a set of “stylistic features” that could be used to categorize paintings, and eventually art objects of different types.
The activity will be performed in collaboration with INRA (FR) and Cambridge University (UK), in the broader framework “ART-ificial Intelligence In Support of Museums”, a grant from Compagnia di San Paolo.
Scadenza validita proposta 30/04/2021
PROPONI LA TUA CANDIDATURA