We have at one point come across giggly girls calling each other “twinnie”, a humbug attempt at expressing love and friendship. Well the guys at Microsoft believe they can help you decipher if in deed your “twinnie” is actually your twin. Microsoft previously allowed us to gauge the age of relatives, family and friends through the widely popular how-old.net. With twinsornot.net , you simply drag images of yourself and the “twinnie” while Microsoft does the work for you.
As experienced with how-old.net, the algorithms are not accurate with twinsornot.net. According to Microsoft, the office light skin Emmanuel Chenze happens to be former President Moi’s twin with a resemblance degree of 37% . While this makes for some interesting questions, there is no semblance whatsoever between the two except for the accents, which I believe were not factored in the algorithms.
Built by the same engineers who built how-old.net , the website also makes use of Microsoft’s Machine learning to assess how similar people in photos look. Microsoft’s Azure Machine Learning Gallery seeks to help people explore more as far as a machine’s ability to learn speech, detect faces and analyze text goes. The website also uses the face API to come up with results as those shown above. The facial recognition bit is to show potential clients how Microsoft’s advanced analytics engines run in the background scouring through data in the cloud (Azure) almost instantly and producing fairly accurate results.
The developer of the site Mat Velloso created the site in just four hours meaning its yet to test on most devices, or even test for bugs besides improving the user interface. We hope that admission from them provides some consolation to both our light skin and the former President.
[…] the same algorithms that power other such solutions we’ve had run-ins with in the past like this one that thought I look like Kenya’s former President Daniel Moi or this one that accurately […]