Day 2 keynote of the Build 2015 was marked by plenty of sessions of code talk. While that was on, Microsoft decided to spice things up a bit with the revelation of a brand new website which can tell how old one looks with the help of a number of algorithms. How-old.net merely requires users to upload a picture of themselves, their family members or their friends. Even someone they search for on Bing can be assessed for their age. It takes seconds before the website shares what age the person looks.
According to the results people have had, it has taken two extremes; it is either too accurate or too inaccurate. It even has a knack for giving variable results for a given person depending on which picture a user uploads for assessment. Suffice it to say, the experience will be a mix bag of flattery and heart break. In other words, you would either want to brag about it or simply shut up and not let anyone know.
The website extracts the gender and age of a person from the uploaded picture. In addition, it accounts for such details as the longitude and latitude of the place where the picture happens to come from. All this data is put up against data uploaded by other users courtesy of Microsoft Azure streaming service. Some other little processes follow in the background before the final result pops up.
This website is a result of casual efforts by Machine Learning team at Microsoft which happened to be toying around with Face detection API of the company. It paired it with Bing Search API to get things going. A casual test soon changed into a search for fifty volunteers to beta test the website which in turn changed to thirty five thousand.