To the untrained eye, spotting an art forgery would be an impossible task, but that’s what art experts are there for, right? Unfortunately it seems that sometimes even experts can get stumped, and that’s where artificial intelligence (AI) comes into play, which is what researchers from Rutgers University and the Atelier for Restoration & Research of Paintings in the Netherlands have developed (via MIT Technology Review).
Given that everyone holds a brush differently and applies different pressure when painting, it’s almost as if brushstrokes are unique to the artist. This is what the researchers relied on by breaking down 300 line drawings from famous artists like Picasso, Matisse, Modigliani, and more. They then fed the strokes into a deep recurrent neural network that identified which were the important strokes to distinguish between the artists.
They also went further by training a new algorithm to look for specific features, like the shape of a line in a stroke. Based on their system, it seems that it was capable of identifying artists 80% of the time. They also commissioned artists to create drawings in a similar style, and the system was able to spot the fakes in every instance.
However there are limitations to this system, and that is it only works when the lines used in paintings are more obvious, but the researchers are planning to test the method on Impressionist works and other 19th-century art.