The Story and Source for TwinsOrNot
- Posted: Jul 08, 2015 at 6:00AM
We are now having some smaller events across the world following Build and I decided to use a few spare hours I’ve had in a hotel in Czech Republic to try another fun site: Twins Or Not
This site takes two photos and will assess how similar the people in these photos are, giving you a score from 0 to a 100. In order to do this, I used the Face API in Project Oxford, a platform of intelligent services. This is just one of the many other amazing features our Machine Learning APIs and platform can offer.
How was it built?
1-I signed up for Azure, Microsoft’s cloud platform
2-Obtained an API key for the Face API, which is a subset of our Machine Learning APIs focused on facial recognition.
3-Downloaded the Face API SDK in .Net which already has a sample that does almost exactly what I needed
4-Adapted the code into an ASP.Net MVC website
5-Published the site to Azure
Total time between the idea and the actual live site: 4 hours (and yes, 4 hours means I haven’t had time to test it with most devices, improve the UI experience or even test so do expect bugs)
Yes, it is that simple. We can create amazing, fun and useful sites and applications one after another, as many as we want. Welcome to the world of Machine Learning.
Source code for www.twinsornot.net v1
A new version of TwinsOrNot is now live and that version doesn't match this source code anymore. This is just for the records and to illustrate how simple these APIs are.
TwinsOrNot is powered by Microsoft Project Oxford, a beta version of a set of APIs, SDKS and services freely available to developers to make their applications more intelligent and engaging. Project Oxford expands upon Microsoft’s evolving portfolio of machine learning APIs adding vision, speech, facial recognition, and language understanding to the set of advanced analysis capabilities. With machine learning technology, it is possible to create sophisticated models that learn from existing data to detect patterns and even predict future behaviors, outcomes, and trends. Developers who want to build breakthrough apps can easily implement machine learning APIs to perform advanced analysis without the expertise of a data scientist.
The popular site How-Old.net is the inspiration for TwinsOrNot.net. Created in two weeks by a team of three Microsoft engineers, How-Old.net quickly became an internet sensation and a brilliant example of these intelligent features. In a similar fashion, TwinsOrNot.net was built by Microsoft developer Mat Velloso in a few hours and was demoed at a //build/tour event to illustrate how anyone can quickly add machine learning intelligence to any application.
TwinsOrNot.net now features a visual refresh and new capabilities including using Bing integration to better compare to your favorite celebrity. We’re also inviting people to help us improve the technology by allowing us to keep their images. These are stripped of identifying information, such as location, and are used to train and improve Microsoft’s computer vision and related services. If people don’t opt-in to help us improve, the images are kept only for the duration of the session and any associated data is automatically deleted within 24 hours.
And using these two very different pictures of me, it seemed to do a pretty good job! :)