Facial recognition methods are all over the place, from safety cameras that attempt to spot criminals to the best way Snapchat finds your face to place bunny ears on it. Computers want a number of knowledge to have the ability to discover ways to acknowledge faces, and a few of it comes from Flickr.
IBM launched a “Diversity in Faces” data set earlier this 12 months, which in a means is arguably factor: a number of early face-recognition algorithms had been skilled on skinny white celebrities, as a result of it’s simple to search out a number of photographs of celebrities. Your knowledge supply impacts what your algorithm is ready to do and perceive, so there are a number of racist, sexist algorithms out there. This dataset goals to assist, by offering photos of faces alongside knowledge concerning the face akin to pores and skin shade.
But most people who uploaded their private snapshots to Flickr most likely didn’t understand that, years down the street, their faces and their buddies’ and households’ faces may very well be used to coach the following massive mega-algorithm. If you utilized a Creative Commons license to your photographs, even a “non commercial” one, you might be on this knowledge set.
NBC studies that IBM says it’s going to take away photos from the information set on the photographer’s or the photographed particular person’s request—however they haven’t made the information set public, so there’s no strategy to see for positive whether or not you’re truly in there. Getting a photograph eliminated gained’t be simple, however if you wish to know whether or not any of yours have been used, you possibly can enter your Flickr username into NBC’s tool here. This isn’t essentially the one knowledge set on the market which may comprise your photograph, however a minimum of there’s a strategy to discover out in case your photographs had been used.