Machine Vision Algorithm May Be the next Art Critic
Gabriel Roşu / 2 years ago
Art historians still find defining art and its creative qualities tricky even today. Art has always been considered something that a human mind can understand and appreciate, but is it really?
A simplified definition of what makes one of the best pieces of art stand out is that they need to have a key element which inspired later artists to use in their own work.
Two researchers at the Rutgers University tend to disagree that humans are the only ones who can judge art. Based on a machine vision algorithm, they proved that paintings can be studied and judged by computers too.
The researchers put it to a test and fed the code a database of about 62,000 pictures of fine art paintings. The results achieved consisted in the computer recognising Monet’s Haystacks at Chailly at sunrise as being one of the most influential paintings in history.
Art critics would agree with the above and state that Rodin’s 1889 sculpture Danaid is not as influential as the machine deems too. But does this really mean a computer can judge fine art now?
The truth is that arguments between critics on determining the most influential pieces of art have been going on for ages now, but this algorithm could prove to be a basis on which they can agree upon.
The algorithm uses visual concepts that analyses both low-key elements, such as colour, texture and simple objects, as well as high-key elements like walking, smiling and so on. A computer then applies the algorithm to a database pool and comes up with the paintings which influenced other authors.
Also, the researchers tell us that the algorithm has bigger potential than just showing a list of paintings. They say that the algorithm can be used in other areas, such as literature, sculpture and even in science.
Thank you MIT Technology Review for providing us with this information
Image courtesy of hdwallpaperpc