Study: Computers Can Spot Painting Forgeries By Checking Brush Strokes

CC0 / / Artificial Intelligence
Artificial Intelligence - Sputnik International
Attention art experts! You may soon be out of a job thanks to a certain group of researchers.

Researchers from Rutgers University and the Atelier for Restoration & Research of Paintings in the Netherlands have published a new paper detailing how their newly developed system can detect forged paintings by analyzing brush strokes.

A persons is seen 05 February 2006 at the entrance to the exhibition of Leonardo da Vinci, scientist and inventor, hosted in the old paper factory of Patras as part of the European Capital of Culture festivities for the year of 2006 - Sputnik International
Leonardo Da Vinci Painting of Christ Sells for Historic $450Mln

Wanting to provide art connoisseurs with another option to detect forgeries, researchers developed a recurrent neural network to learn what brush strokes and specific features correlated with certain artists. In all, the system broke down paintings from artists like Pablo Picasso, Henri Matisse and Egon Schiele into 80,000 brush strokes.

After breaking down the brush strokes, the algorithm was then taught to identify the curves, waves and weight of the lines in order to detect how much pressure the artist was putting on the paintbrush when they were painting on the canvas.

Putting the forgery-detection system to the test, researchers then enlisted the help of fellow artists to see if they could recreate a painting that would surpass the AI's abilities — it did not disappoint. Every single time researchers put the real and fake paintings against one another the system was able to identify the forgery.

Pablo Picasso suns himself on a boat on the beach at Golfe Juan in Vallauris on the French Riviera on March 10, 1948. - Sputnik International
Picasso Painting Stolen by Nazis Fetches $45 Million at New York Auction
"The experiment shows that the proposed methodology can classify individual strokes with accuracy 70%-90% and aggregate over drawings with accuracy above 80%," the study reads, "with accuracy 100% for detecting fakes in most settings."

However, since the system bases its findings on brush stroke lines it is unable to detect forgeries when the brush strokes have faded out or the paintings are simply too old.

In the next phase of testing, researchers will be using Impressionist artworks and other 19th century pieces to further validate their efforts.

The present methods officials typically use are infrared spectroscopy, radiometric dating, gas chromatography, or a combination of all three to detect fake paintings, Technology Review reported.

The team will be presenting their findings at the 32nd AAAI Conference on Artificial Intelligence in February 2018.

To participate in the discussion
log in or register
Заголовок открываемого материала