Researchers think algorithms can improve your trash selfies
Whether we take selfies to match our own perception of what we think we look like or trying to understand how others perceive us, there’s no doubt that the self portrait is the defining photographic trend of our time. Chances are, we could all use a little help in upping our selfie game, which is where new research and the resulting algorithm-powered smartphone app comes in.
Computer scientists at the University of Waterloo in Canada asked thousands of Mechanical Turk workers to rate artificial selfies to find the best choice across three dimensions: lighting direction, face size, and face position. They had workers rate the selfies and then created an app that acts as a “director” for selfies, helping you get your best possible self-portrait. The researchers then had real people take photos with and without the app and submitted the selfies to raters from Mechanical Turk. They found a 26 percent improvement in the ratings of self-portraits taken with the app than those taken without direction.
The study’s co-authors Qifan Li and Dan Vogel think the research can go even further. “This is just the beginning of what is possible,” Vogel said in a statement. “We can expand the variables to include variables aspects such as hairstyle, types of smile or even the outfit you wear. When it comes to teaching people to take better selfies, the sky’s the limit.”
While the app the researchers created isn’t commercially available, it’s likely we’ll start to see them out in the wild soon. We can all use a little direction when it comes to taking better selfies.
Source: University of Waterloo