Microsoft Garage launches machine learning-based tech

Nov 9, 2015 13:28 GMT  ·  By

You probably remember Microsoft’s How-Old.net service that employed machine learning technology to try to determine your age using just a simple photo.

Microsoft Garage, the company’s division in charge of building experimental projects that could at some point become the foundation of much more advanced products, has decided to go one step further with another similar service, this time trying to guess the length of your facial hair. And if you don’t have any, to put a mustache on your face.

The new project is called MyMoustache.net (via WMPU) and is based on the very same idea as the age-guessing game: you only have to upload your photo to the service (Microsoft guarantees that no pictures belonging to users are stored on their servers) and then let it do its magic to calculate the length of your facial hair and compare it to the others'.

How do I look with facial hair?

If you don’t have any facial hair, as is the case of our beloved Julie Larson-Green we used for the test, the service automatically picks a moustache that’s supposed to match your face (some improvements here are still needed, as you can see).

“We started with a simple question, ‘What if we could use machine learning technology to detect the length of facial hair?’ Many hallway jokes, brainstorming sessions, and mockups later, our early ideas have been realized far beyond our initial expectations,” Microsoft says in the description of the service that’s posted on the official website.

Obviously, do not expect this service to be 100 percent accurate, but it’s further proof that machine learning technologies can serve as the foundation of an infinite array of products, and Microsoft’s Project Oxford Face API is a set of tools that can definitely help in this regard.

Just like How-Old.net and TwinsOrNot.net, expect MyMoustache.net to become viral in just a few days.

MyMoustache.net in action (4 Images)

MyMoustache.net at work
MyMoustache.net at workMyMoustache.net at work
+1more