Alexa may not seem very visual (after all, she doesn't have a screen) but the home assistant can now turn visuals into stories thanks to a new startup.
A photographic memory is something few can claim — but soon Amazon’s personal assistant Alexa may be able to remember vivid details from your life, or even a relative’s life. Mylestone, a startup that just landed $2.5 million in funding, is an Alexa app that gives the assistant the ability to tell personal stories from your past, or your family history, based on photos.
Currently, Mylestone feeds Alexa that information based on photos that are uploaded into the system. But, the start-up hopes to eventually allow the platform to link to images shared across social media platforms to automatically add new images for Alexa to retell.
Those images are interpreted using artificial intelligence — and, for now, a team of actual people. The computer uses object recognition as well as facial recognition to identify the objects and people in the images. And like a recent Facebook update that recognizes scenes like the Golden Gate Bridge or the Eiffel Tower, the system also recognizes landmarks. Mylestone staff also help to add the details that artificial intelligence hasn’t learned to pick up on yet.
With that information in the system, users can use commands like “Alexa, tell me a memory about Mom,” and, using those photos, the program will turns images (or, more accurately, series of images) into a story. Aspects like the date, found inside a photograph’s metadata, and objects and people in the scene are turned into a story.
According to the sample stories on the Mylestone website, Alexa can also combine the details in the photos with general, Google-accessible information; for example, by adding a quip about a certain landmark that’s recognized in a scene.
Since the platform can also read audio and video files, users can even ask the system something like, “Alexa, have Grandma tell us the Thanksgiving prayer” and hear specific voices besides Alexa’s pulled from those files.
The program is in its infancy, but the team behind it expects to advance the program using automated uploads tied to social media, as well as by enhancing the capabilities of the artificial intelligence that recognizes the photos.