April 30, 2015
As I’m sure you’re aware, the second day of the Build Developer Conference is winding down There’s been a tremendous amount of hype, demonstration, and showcasing up to this point, but through all that noise today cut one clear chord – Microsoft.
However, Microsoft’s flex was a bit more technical. In effect it was a demonstration of machine learning – how-old.net being the platform it operates through. It sifts through metadata attached to photos to tell you what the age and gender of the person(s) in the photo is.
Now, I’d wager that the majority of you reading this have already used the site, so you know that you can either use a photo provided or upload your own. Sometimes how-old.net nails it square on the head, but sometimes it lights up a room with laughter because it’s so far off.
But remember, this is a debut of Microsoft’s machine learning platforms, and they’ll steadily improve in accuracy as more photos are uploaded and processed.
If it looks simple on the surface, that’s only because the team has made it that way. In reality there’s a hell of a lot going on behind the scenes.
Corom Thompson and Santosh Balasubramanian, both Engineers in Information Management and Machine Learning at Microsoft, wrote a blog post describing the genesis of the website as well as the technical aspects that make it work.
“We were expecting perhaps 50 users for a test but – in the end – got over 35,000 users and saw the whole thing unfold in real time,” reads the post. “We also got real time insights to learn more about how people were using this tool. For instance, we assumed that folks would not want to upload their own pictures but would prefer to select from pre-canned images such as what they found online. But we what we found out was that over half the pictures analyzed were of people who had uploaded their own images.”
According to Thompson and Balasubramanian, the platform works by collecting and analyzing data obtained from the photo in real time. Note, the picture itself isn’t saved, only the metadata attached to the picture is extracted and saved. It’s then passed through not one, but two other analytic tools in the Microsoft Azure streaming services, producing the finished image with an age and gender stamp in real time.
The authors are using this as a great opportunity to showcase the Azure services and the APIs available in the ML Gallery, which by extension were showcased to the world at the conference today.
“We hope you have fun with it and are inspired to create your own solutions using Azure services and the APIs available in the ML Gallery,” finishes the blog post.
Here are some people I thought would be fun to run through the platform:
Photo Credit: Air Herald
Photo Credit: Salon
Photo Credit: Fandango
Photo Credit: Wikipedia
Photo Credit: peoplequiz.com
Lead Image Credit: how-old.net homepage
Did you like this article?
Get more delivered to your inbox just like it!