As the leading private company offering brushstroke AI authentication, Art Recognition, based in Zurich, works by designing a deep convolutional neural network (DCNN) specific to the artist. The brushstroke AI was trained on a high-quality dataset containing images of almost all the paintings ever created by Raphael and a contrast set of fakes, comparables, and contemporaries. The system then identified feature series, such as the artist’s brushstroke and variations in color and high-level composition elements.

The AI reviewed 16 individual segments of the Flaget Madonna and the painting as a whole. While the analysis showed that the painting was not entirely by Raphael, both the faces of the Madonna and Child were confirmed as 96.57% and 96.24% attribution to Raphael.

“Through brushstroke artificial intelligence, we offer objectivity and accessibility to clients, which has been missing the field of art evaluation for many years. Art history, provenance, chemical analysis, and other methods are all critical to fully understanding an artwork. Still, attribution decisions should not be left solely to the subjective human expert’s eye,” said Dr. Carina Popovici, CEO of Art Recognition.

“With brushstroke AI making advancements in the art world, it’s exciting to see Art Recognition use this technology to authenticate paintings,” added Hikari Senju, Founder, and CEO at Omneky, which utilizes state-of-the-art deep learning to empower businesses to grow. “Growing up, I remember going to art museums with my dad, who is an artist, and him pointing out characteristics of each artwork and what made them significant. With the advent of computer vision, art, and design can be quantified for the first time. Never before have we had the power to attribute artwork with the support of AI to confirm an artist's technique with data. We are in an exciting period in technological design with wide implications of this innovation in various fields.”