Style Painting Classifier Based on Horn Clauses and Explanations (SHE)

Vicent Costa*, Pilar Dellunde, Zoe Falomir

*Corresponding author for this work

Research output: Chapter in BookChapterResearchpeer-review

1 Citation (Scopus)

Abstract

This paper presents an style painting classifier, based on Horn clauses in rational Pavelka logic and qualitative colour descriptors, that provides explanations (SHE). A fuzzy representation of colour traits for Baroque, Impressionism and Post-Impressionism art styles is introduced. The SHE-algorithm has been implemented in Swi-Prolog and tested on the 90 paintings in QArt-Dataset. the Baroque style was classified in 90% of accuracy. The general accuracy for all the art styles was 73.3%. SHE provided explanations of right classifications, and also of outliers by giving a second option depending on the membership degree of the painting to a certain style.

Original languageEnglish
Title of host publicationArtificial Intelligence Research and Development
Subtitle of host publicationCurrent Challenges, New Trends and Applications
EditorsKarina Gibert, Zoe Falomir, Enric Plaza
PublisherIOS Press
Pages37-46
Number of pages10
ISBN (Print)9781614999171
DOIs
Publication statusPublished - 2018

Publication series

NameFrontiers in Artificial Intelligence and Applications
Volume308
ISSN (Print)0922-6389

Keywords

  • art
  • classifier
  • explainable AI
  • Horn clause
  • logic programming
  • qualitative colour

Fingerprint

Dive into the research topics of 'Style Painting Classifier Based on Horn Clauses and Explanations (SHE)'. Together they form a unique fingerprint.

Cite this