May 24, 2011

Photos Of People More Memorable Than Landscape Photos

MIT neuroscientists have found that the most memorable photos taken are those that contain people, followed by static indoor scenes and human scale objects. 

The researchers said that landscapes are the most forgettable images in most cases.

"Pleasantness and memorability are not the same," MIT graduate student Phillip Isola, one of the lead authors of the paper, said in a statement. 

This study is the first to model what makes an image memorable.

"People did not think it was possible to find anything consistent," Aude Oliva, associate professor of cognitive science and a senior author of the paper, said in a statement.

However, the neuroscientists found consistency among hundreds of people who participated in the memory experiments.

The researchers are developing a computer algorithm that ranks images based on memorability.  Oliva said this algorithm could potentially be useful to graphic designers, photo editors, or anyone trying to decide which of their vacation photos to post on Facebook.

Researchers built a collection of about 10,000 images of interior-design photos, nature scenes, streetscapes and others. 

Human subjects in the study were shown a series of images, some of which were repeated.  They pressed a key on their keyboard to indicate an image they said they had already seen.

Different research subjects produced similar memorability ratings.

"There are always differences between observers, but on average, there is very high consistency," Oliva said in a press release.

The team made "memorability maps" of each image by asking people to label all the objects in the images. 

Alexei Efros, associated professor of computer science at Carnegie Mellon University, said the study offers a unique way to characterize images.

"There has been a lot of work in trying to understand what makes an image interesting, or appealing, or what makes people like a particular image. But all of those questions are really hard to answer," Efros, who was not involved in this research, said in a press release. "What [the MIT researchers] did was basically approach the problem from a very scientific point of view and say that one thing we can measure is memorability."

The researchers generated an algorithm that can predict memorability of images the computational model as not "seen" before.  MIT said an algorithm like this could be used by book publishers to evaluate cover art, or news editors looking for the most memorable photograph to feature on their website.

The team is doing a follow-up study to test longer-term memorability of images as well.

The paper will be presented at the IEEE Conference on Computer Vision and Pattern Recognition taking place June 20-25 in Colorado Springs.


Image Caption: Can you guess which of these images would be the most memorable? Give up? It's the top left and bottom right ones. Images courtesy of the Oliva and Torralba labs


On the Net: