Yahoo is announcing the release of a brand new function in Flickr: the potential to search for photographs which can be visually similar to a single given image. Strolling this kind of searches is as simple as hovering over an picture you’re interested by inside seek results, clicking at the three dots, and deciding on “search for similar pictures.”
Flickr makes use of deep neural networks to examine the pictures as they’re uploaded to the server. In preference to only showing pix containing the identical object, Flickr’s visually comparable photo seek — internally known as the “similarity pivot” — fetches pics that have sure definite connections, like shape, color, or fashionable categories. It’s similar to what facebook these days released last month in which users can search for pix based totally on their content, without requiring the uploader to manually add tags.
Pinterest’s Lens characteristic does some thing like that once it finds pins which are visually similar to what human beings take photos of with their cellphone cameras. but Flickr’s facts set is essentially one of a kind; it’s usually user-generated content in place of curated collections of product photography and different expert content material. So it doesn’t overlap. On the equal time, it’s related due to the fact Flickr’s new function, like Pinterest’s, provides a distinct way to exhibit available content material for customers.This isn’t the primary time Yahoo has enhanced Flickr with deep learning, a kind of synthetic intelligence that includes training artificial neural networks on information and then getting them to make inferences approximately new information. For one thing, Yahoo makes use of it to determine if pictures aren’t safe for work (NSFW).
Flickr won’t be the primary region you visit while seeking to look for pix, but it’s miles a useful device for finding creative Commons-licensed snap shots that healthy your exact visual wishes. The Similiarity Pivot device is available now.