Microsoft’s age guessing robot uses machine learning to analyze your uploaded photos. With a new feature users can now have a guess themselves and help fine tuning the algorithms.
In April, Microsoft opened How-old.net to introduce Azure Services to people while entertaining them. The website lets you upload a personal photo and then has a guess on gender and age.
With the new feature announced on the Microsoft Blog, the website adds a small game of age-guessing.
After you upload your photo or choose a photo from Bing search, the algorithm guesses your age like before. In addition now you also can ask other active users of How-old.net to have a guess.
The new option is shown while the How-To.Net robot is analyzing an uploaded photo. When users join in, the machine algorithm benefits from additional data to sharpen its future estimates.
Gender and age guessing is nothing new to Microsoft
The website uses face detection routines from the company’s Project Oxford, Machine Learning API’s to apply real time streaming analytics and real time Business Intelligence.
The key components of the application are:
- Age and gender extraction from photos
- Real time insights over the above data
- Creating real time dashboards to show results
How-old.net, which started as a small experiment for a Build2015 Developer Conference, quickly became a worldwide hit. Within hours, over 210k images were submitted and the site reached 35k users from all over the world.
Source: Microsoft Blog