Creating avatars. Who’s got time for it?! Computers, that’s who. You’ll ne’er have to waste another second choosing your hair style, skin tone or facial hair length if this analysis from Facebook finds its method into product form.
In a paper (PDF) presented at the International Conference on computer Vision, Lior Wolf et al. show however they created a machine learning system that makes the most effective possible match of your real face to 1 in a custom emoji generator.
You may be thinking: Wait, didn’t Google do that earlier within the year? yes, kind of. but there’s a critical difference. Google’s version, while cool, used humans to rate and describe varied options found in common among various faces: curly hair, nose types, eye shapes. These were then illustrated (quite well, I thought) as representations of that specific feature.
Essentially, the pc appearance for the tell-tale signs of a feature like freckles, then grabs the corresponding piece of art from its database. It works, but it’s mostly reliant on human input for defining the features.
Facebook’s approach was totally different. the concept being pursued was a system that actually makes the most effective possible representation of a given face, using whatever tools it’s at hand. so whether it’s emoji, Bitmoji (shudder), Mii, a VR face generator or anything, it may still accomplish its task. To paraphrase the researchers, humans do it all the time, so why not AI?
The system accomplishes this (to some degree) by judging each the face and therefore the generated representation by the same analysis and have identification algorithm, as if they were merely 2 footage of the same person. once the resulting numbers generated by the 2 are as close as they seem likely to get, which means the 2 are visually like a sufficient degree. (At some point with these cartoon faces it isn’t going to get far better.)
What’s great about this method is that because it isn’t tied to any specific avatar type, it works (theoretically) on any of them. As long as there are good representations and bad ones, the system will match them with the actual face and figure out that is which.
Facebook may use this info for many helpful functions — perhaps most immediately a bespoke emoji system. It may even update automatically once you put up an image with a new haircut or trimmed beard. but the avatar-matching work could even be done for different sites — sign into whatever VR game with Facebook and have it instantly create a convincing version of yourself. and plenty of people out there surely wouldn’t mind if, at the very least, their emoji defaulted to their actual skin color instead of yellow.
The full paper is pretty technical, but it was presented at an AI conference, so that’s to be expected.
Rumored voice-controlled TiVo’s DVR pops up at Amazon, Best Buy
Source:techcrunch.com
