Finding the perfect pet with machine vision technology

An app called PetMatch from startup Superfish utilizes machine vision algorithms for image recognition purposes in order to help you find your dream pet, or find a lookalike to a previous pet, in your area.

The program allows a user to upload a picture of a dog or cat, then uses its proprietary image recognition technology to scan the photo register certain details of the animal’s mouth, distance between its eyes, its color, as well as its general size and shape. The app then algorithmically searches through its database of photos of animals that are up for adoption, to find a potential fit.

For images to be searchable, explains Superfish, they require metadata and text tags, but most images out there do not have this. The visual search that Superfish utilizes "liberates images from this dependence," by using algorithms to analyze photos in detail. Superfish co-founder and CEO Adi Pinhas told TechCrunch that his goal is “to change the way people use the camera in their mobile.”

Of course, the app will become more useful as the database of photos continues to grow. In an article on CNN, the author notes that they performed a test using a photo of a gray and white Scottish Fold Cat in Atlanta, and it returned matches of cats with similar color in Jefferson, Georgia; Knoxville, Tennessee; Huntsville, Alabama; Charlotte, North Carolina, and other neighboring states. While some of the cats looked a bit like Cash, most did not.

While the app’s overall success and ability to produce similar looking potential pets may hinge on the amount of photos and adoptable pets available, the idea is a novel one. Where else have you seen machine vision technologies used outside of their normal scope? Let us know in the comments below or email me here. Who knows, maybe your idea will be our next blog.

View other items from our July 7 e-newsletter.