Friday, February 12, 2016

Software : Microsoft's facial recognition AI can sniff out your dog's breed

Software : Microsoft's facial recognition AI can sniff out your dog's breed


Microsoft's facial recognition AI can sniff out your dog's breed

Posted:

Microsoft's facial recognition AI can sniff out your dog's breed

Microsoft previously teamed up its facial recognition and artificial intelligence teams to figure out how old people are, with mixed results. The firm's latest app, on the other hand, aims to show off its machine learning chops by discerning your dog's breed.

Microsoft's experimental lab Garage has launched Fetch!, an app that's powered by the company's Project Oxford AI platform and machine learning system that can determine the breed of a canine from a photo on your phone's camera roll or a new snap.

The breed-recognition technology, which was first developed two years ago, has been turned into an app to show that object recognition could be "extraordinary, fun and surprising."

"We wanted to show that object recognition is something anyone could understand and interact with," said a Microsoft Research development director and project leader Mitch Goldberg in a blog post.

Experimental Garage

Fetch! won't get it right every time - it confused my cat for a miniature Schnauzer because of the way she was sitting. But the app is impressive in that it guesses the percentage of the breed match, too, as well as providing information on the breed type.

But perhaps more fun is that it will guess the type of dog a person would be if you snap a picture of yourself, your friends or family.

Fetch!, currently available for iOS, is the latest in a range of apps and experiments coming out of Microsoft Garage and Project Oxford. Just last month, Microsoft launched the playful alarm app Mimic for Android that was also developed through Garage and used to show off some of Project Oxford's machine learning APIs.

You can check out Fetch! in action below:

YouTube : https://www.youtube.com/watch?v=JWMqFp1BPus

No comments:

Post a Comment