Have you ever seen or met a dog and were unsure about the breed? It can be more difficult than you would think. If you don't have an idea of what breed it might be, you will probably spend a good amount of time flipping through dog pictures, and even then many breeds can have similar features that makes this even more difficult. To take the guesswork out of this, Microsoft has released a new app, cleverly named Fetch!, which utilizes deep neural networks to identify dogs by breed, as well as characteristics unique to them.
This technology should sound familiar, in fact Fetch! makes use of the same Project Oxford APIs we have seen used by Microsoft Garage apps in the past like Mimicker, as well as websites like Twins or Not and How-Old.net. For those who aren't familiar, Project Oxford provides access to a number of vision, speech, and language APIs that leverage the power of machine learning on a platform that is accessible to smaller companies who may not be able usually afford developing such a program on their own.
From within the app, you can snap photos of dogs and submit them to be analyzed. After producing a match, the app can then provide a history of the dog breed as well as some notable characteristics. You can also scroll through a list of dog breeds if you just want to learn more about them. For some extra fun, you can snap a photo of you or your friend and the app will identify the closest dog breed your physical characteristics match and then share your fun findings with friends.
While the app is more for having fun with your friends guessing dog breeds, it could also be a handy tool for first time dog adopters looking to find the perfect pet. Identifying breeds, characteristics, and whether or not the pet will be right for you and your family are just a couple of the struggles and considerations you may encounter on such a journey.