Two computer scientists at the University of Waterloo have created an algorithm that gives smartphone users new tools to work with for capturing significantly better selfie photos. Unlike other camera applications that improve images through post-processing methods, the idea behind this particular selfie application is to provide users with instructions on how to improve their selfie before it is captured.
The application uses an algorithm developed by Daniel Vogel, a professor of computer science at the Waterloo University, and his former Master’s student Qifan Li. The algorithm is designed to guide users on how to improve three main compositional features of portrait photography including lighting direction, face position, and face size. The instructions are delivered through the camera app’s user interface in a somewhat similar fashion to panoramic shots, where on-screen indicators show the direction in which the camera should move. However, this is an entirely different endeavor as the main goal here is to find the perfect photography conditions rather than guiding the user’s hands on a horizontal plane for a wide panoramic shot. The algorithm was created with the help of an online crowdsourcing service, where the researchers have presented a gallery of virtual selfie photos to thousands of people. The participants voted for which selfie shots they felt looked best, and based on that data, the computer scientists have mathematically modeled the patterns of votes in order to create the aforementioned algorithm. Afterward, the researchers have created new selfie shots using a standard camera application and the app powered by their algorithm, and found that the selfies captured with their application presented a 26 percent improvement, according to additional online feedback from participants.
Evidently, the algorithm could technically become increasingly better at determining how to improve the aforementioned compositional features of portrait photography by observing more voting patterns for a wider variety of selfie shots. “This is just the beginning,” according to Vogel, who believes that the algorithm could also further expand to include other variables such as hairstyles, facial expressions like smiling, and even outfits. Refer to the YouTube video below to see how this unconventional creation works in practice.
The post Waterloo Researchers Have Created ‘A Perfect Selfie’ App appeared first on AndroidHeadlines.com |.
【Top 10 Malaysia & Singapore Most Beautiful Girls】Have you follow?
