Are Computers Able to Understand Art?

Art is deeply connected with subjectiveness and human perception has a great insight in understanding and interpreting art. In this chapter we follow the attempts of computer vision researchers to teach computers ways of understanding artistic paintings. Can computers “see” past the abstraction and can they understand the deeper, intuitive levels in art? Are digital tools able to help in such a subjective and delicate domain? We will discuss methods that analyze paintings for the recognition of artistic style and artistic influences, which can help the art historians and curators from art museums. We will delve into algorithms for recognizing the subject in a painting and how a computer can get past the abstraction of art. Furthermore, we will analyze how the computer vision researchers tried to overcome the limited data by teaching computers to paint new pieces based on known art masterpieces. Based on these techniques, computer vision can help indexing big datasets of images or can provide the normal, untrained viewer of paintings with new art that they might enjoy or help them discover how to perceive the artworks using hints from color palette or attention gathering details.