Microsoft Garage is ofcourse, one of the more inquisitive branches within the Redmond giant. The Garage is Microsoft’s testing lab where employees design and develop products and services which have nothing to do with the company’s main goals. On Friday, Microsoft, via the Garage announced Fetch!, an iOS app that attempts to help you figure out what breed your dog belongs too.
The app obviously uses deep learning technology based on artificial neural networks which record and store data in real-time. This is one of the most widely used and better approaches to artificial intelligence used in big data storage etc.
Fetch! isn’t Microsoft’s first app in this domain though. It’s just the next step in what Microsoft believes is a series of releases that try to make understanding the complexities of machine learning more accessible to the mainstream user. We saw the company release many apps and websites last year. We saw a site that could guess your age, a site that will tell you if two people are twins and a lot more. All these were results of years of work by researchers in Microsoft Garage.
Now let’s see what all this dogging is about.
The app released by Microsoft today can use a photograph from your gallery or utilize your iPhone camera to take a snapshot. The pic is then scanned by the app and matched for features over a neural network containing data which is updated in real-time. This means the app learns from each operation. After matching the picture with the database, the services attempts to determine what type of dog is in the picture.
Accuracy is a hit or miss as it depends on many factors like the distance to the dog, lighting conditions and a lot more. But the Redmond giant is optimistic about the fact that as usage increases and the neural networks develop better, the apps performance will skyrocket.
If the app isn’t able to point out exactly which breed of a dog you have, it shows you possible answers. It can also detect whether or not a dog is present in a picture.
There was an interest in creating a framework that would allow you to take a domain – in our case, dogs – and recognize numerous classes, such as breeds. We were interested in enabling an app to allow you to make object recognition extraordinary, fun and surprising,
says Mitch Goldberg, a development director at Microsoft Research whose Cambridge, UK based team built the experience.
We wanted to bring artificial intelligence to the canine world. We wanted to show that object recognition is something anyone could understand and interact with.
Well that’s for the technical details. Now let’s talk about the fun part.
The app can also bring out the dog in you. Or your friends. Or that creepy guy who is sitting beside you in the bus. This Easter egg feature can scan a human picture and determine what dog breed the person is most closely like.
The service was under development since 2014. A demonstration of the technology behind Fetch! appeared on stage at the 15th annual Microsoft Research Faculty Summit.
I see Fetch! as a great example of how a carefully crafted user experience can enhance the power of machine learning,
says Ken Woodberry, deputy managing director at Microsoft’s research labs in Cambridge.