Search

Showing posts with label artwork. Show all posts
Showing posts with label artwork. Show all posts

When computer algorithms for artworks can match art historians, accuracy will match precision


Title picture
Biker and Son, Poster art. credit: Flickr.com, Moriza
When art historians evaluate an artwork, they ask and answer questions such as the time and place where the artwork was done, what school of art the artist represents as well as his influences, and the artists his work has influenced. Asking a computer using an algorithm to do this would be very difficult. This is a field where human intelligence is very much required. Yet, computers and computer algorithms can still help us novices evaluate a work of art.

Algorithms have been developed that demonstrates computer ability to perceive and understand art the same way expert art critics would. The algorithms work on the composition of colors and easily measurable artistic qualities of an artwork.

Computers as art historians: Good job done!

In a recently published experiment in the ACM Journal of Computing and Cultural Heritage, using approximately 1,000 paintings of 34 well-known artists,
Advert
205747_Holiday 30% off card and calendars
researchers showed that a computer was clearly able to identify different art styles and even little nuances between art schools using only visual content. The results obtained were largely in agreement with the perception of art historians and dwarfed what one can obtain from untrained humans in the analysis of artworks.

A similar algorithm that was designed to place artworks in particular artistic periods using artificial vision algorithms also gave outstanding results. The computers were able to “understand” images and also differentiate between artistic styles based on low-level pictorial information.

Low-level pictorial information encompasses aspects such as brush thickness, the type of material and the composition of the palette of color. Humans on the other hand perform art analysis more abstractly. They employ medium-level information which differentiates between objects and scenes in a picture and the type of painting, and high-level information that takes the historical context and artist’s knowledge along with artistic trends into consideration.

Computer algorithms for arts not yet fully explored.

The future is stillborn as to the scope and extent visual algorithms can be used in evaluating works of art. At least, when it involves broad artistic differentiation, usually amongst art schools and periods, computer algorithms can beat an untrained human in arts analysis but fall short in precisely dating art periods or styles. According to Miquel Feixas, one of the authors of the study that was published in the ACM journal, “it will never be possible to precisely determine mathematically an artistic period or to measure the human response to a work of art, but we can look for trends.”

This is not the only technique that has been expended towards paintings. Physicists sometimes use ion beams to determine paintings and painting authenticity. This is the first time though a machine has gone one step better than humans, whether trained or untrained, in evaluating and analyzing works of art. It is hoped that this work will also open the way for further work in developing image viewing and analysis tools, classifying and searching for collections in museums, creating public information and entertainment equipment and to better understand the interaction between humans, computers and works of art. Digital technology will not replace art historians at all, just as software has not replaced human mathematicians, but with digital technology, art historians have an arsenal that makes their work stand out with more precise accuracy.

Advert


follow me on twitter, @emeka_david or be a friend on facebook, odimegwu david

Matched content