Can’t decide which photos to toss? EyeEm’s AI helps you find the keepers

EyeEm app
EyeEm

Think a computer can pick out your best photos? Artificial intelligence photo app EyeEm thinks it can. On June 22, EyeEm launched an update that now suggests photos from the camera roll to upload.

The feature, EyeEm Selects, is launching first to Android users. It will come to iOS in a future update, but a similar tool already exists in a separate EyeEm iOS app, called The Roll. The update builds on EyeEm’s existing AI photo network, EyeEm Vision, and introduces a way for the computer to automatically select the best shots.

To train the computer, a team of human photo curators collected “good” photos as the computer watched what was selected and what wasn’t. As the photo editors sorted through millions of photos, the computer learned a number of different qualities that makes a photo more likely to be selected.

Unlike other attempts to teach computers to recognize good photography, the system isn’t based on technical aspects, like whether the shot is sharp or if it uses the Rule of Thirds. “In an artistic medium like photography, photographers constantly explore and innovate,” said Appu Shaji, EyeEm’s head of research and development. “Images that deviate from the established rules are often the ones that evoke strong aesthetics. For this reason, we purposely dissuaded the photo curators from deconstructing the technical aspects and encouraged them to use their innate visual sense and judgement. We have thus developed our own aesthetic criteria with which to train our dataset.”

With the app update, users will see “EyeEm Selects” located above the camera roll. Tapping that option allows the system to search for the best photos. The entire process isn’t cloud-based, so the photos never leave the device and no one sees any of the files until they are uploaded, EyeEm says.

The feature is designed to help users find the hidden gems that already exist in the camera roll, but that simply weren’t shared right away. Machine learning and artificial intelligence is now a major part of photo apps. Besides EyeEm, companies – big and small – like Google, Apple, Shutterstock, Dreamstime, Adobe, and others, are building AI into their software to handle everything from auto-tagging, face detection, and even best moments.

EyeEm, a Berlin-based company founded in 2011, is a mix of a social sharing and stock photos. The app offers tools like auto keywording (powered by the same EyeEm Vision AI system) and a sales agreement with Adobe Stock.