What Dog Breed is That? Let AI “fetch” it for you!

Recently released AI app identifies dog breed information from pictures and mixes some fun too.

Ever wished for on-demand dog education to help earn social prestige or even just throw an ice-breaker (and make new friends)? Wish granted! Fetching the dog breed information has never been easier.

Microsoft Garage recently launched an app called “Fetch!” that can identify a dog’s breed from pictures. The app uses Artificial Intelligence (AI) for classifying the image of a dog to its knowledge base of over a 100 popular pure breeds.
"It's about bringing artificial intelligence to the canine world," Mitch Goldberg, development director at Microsoft Research said. "Computers have been able to hear and talk, and now they can see."

Once it identifies the dog breed, the app also provides interesting details and characteristics of that breed. This can also help you in dog research to identify what breeds will be best suited to your needs and lifestyle. what-dog

The app runs in continual learning mode, aiming to get better over time. Currently, it is not great at recognizing mixed breeds, and developers are already working on that improvement.

The app is available for free on the Apple App Store as well as on web at What-Dog.net.

The app also has a fun side to it. In the fun mode, you can find out what dog breeds you and your friends are. Goldberg said that if users " take a picture of a person, it'll kick into its hidden fun mode. And in a playful way, it'll communicate to you not only what type of dog it thinks you are, but also why."
Microsoft Garage, self-described as “an outlet for experimental projects for you to try”, is a Microsoft project lab where employees work on creative projects (during off-duty times) often unrelated to their primary job responsibilities. The initiative was publically announced in October 2014. In the recent past, Microsoft has released several fun applications of Artificial Intelligence, such as TwinsOrNot.net, MyMoustache.net and the viral hit How-Old.net.