Hashtagging pictures of your #pitbull on Instagram is accomplishing more than just connecting you to other dog lovers. Facebook announced Wednesday that it's been using publicly available, hashtagged photos to train computer vision models -- and it's achieved breakthrough results. Computer vision models typically rely almost entirely on hand-curated, human-labeled data sets. This makes it the biggest limiting factor in computer vision, Facebook CTO Mike Schroepfer said on Day 2 of the F8 developer conference in San Jose. To address this, Facebook has instead trained models with a set of 3.5 billion publicly available images and 17,000 hashtags.
Chrome: Tons of photography apps, like Google Photos and Apple Photos, try and automatically make sense of objects in your photos and add automated tags, and it turns out Facebook does that too, even though you'd never know it. Show Facebook Computer Vision Tags is a Chrome extension that reveals all those tags. Facebook uses software to look at the images you upload, then adds tags based on what it sees. The only aspect of this software you'll see is the facial recognition feature that suggests a face in an image, but Facebook also tags photos with data like how many people are in a photo, whether it's indoors or outdoors, and more. This is useful for a number of reasons, but it's mainly used for blind users who might depend on screen readers to understand an image.
As originally announced in June, game developer Blizzard Entertainment and social media powerhouse Facebook have agreed to a deal that enables FB users to stream their Blizzard gameplay over Facebook Live. And, starting Friday (hey, that's today!), users will actually be able to. The service is currently limited to PC-gamers in the Americas, Southeast Asia, Australia and New Zealand, though Blizzard is working to expand to other platforms and regions. In order to enable streaming, simply connect your Battle.net