Analysis of Dutch Master Paintings with Convolutional Neural Networks
This addresses the problem of art authentication and attribution for art historians and conservators, representing a domain-specific application of existing methods to new data.
The paper tackled the problem of authenticating and attributing Dutch Master paintings by using convolutional neural networks trained on the artist's works and comparable works of others, resulting in the ability to identify forgeries, provide attributions, and detect mixed authorship within paintings with classification probabilities.
Trained on the works of an artist under study and visually comparable works of other artists, convolutional neural networks can identify forgeries and provide attributions. They can also assign classification probabilities within a painting, revealing mixed authorship and identifying regions painted by different hands.