You know that moment when you’re telling a great story, the one about the time your friend carved Aaron Rodgers’ likeness out of cheese and then married it, but in the moment, you can’t remember where to find the photo evidence on Facebook? Your storytelling skills may be vastly improved now that Facebook will let users search among their friends’ photos.
Facebook announced today that its AI can now do things like “recognize what’s in an image, what type of scene it is, if it’s a well-known landmark, and so on.”
The company says it’s already made advances in deep learning that allows it to classify images using questions like “What is in the image?” and “Where are the objects?” New techniques can now detect and segment the objects in a given image, and identifiy objects and scenes, allowing the AI to attach “even more meaning to the photo.”
The new search system leverages this image understanding to sort through all that information and surface the most relevant photos, Facebook explains.
“In other words, in a search for ‘black shirt photo,’ the system can “see” whether there is a black shirt in the photo and search based on that, even if the photo wasn’t tagged with that information,” Facebook says.
Or maybe, “eating pizza with cat and wine picture,” as another totally random example.
That way, users don’t have to rely on remembering the caption that went with the photo, the people tagged in it, or the album’s name where it appeared.
Will this system work as intended, or will there be some irrelevant (and likely inappropriate) results? That remains to be seen, but Facebook says that to ensure search results are relevant to the query, its team used deep learning techniques to get to the semantic meaning of billions of photos, and thus, better rank photos.
(via The Verge)
by Mary Beth Quirk via Consumerist