Animal nose lines, like human fingerprints, do not change with age, and even identical twins hold different lines, making them suitable for biometrics. A group of Japanese companies have teamed up to conduct a nose print recognition experiment using photos of dogs’ noses and to develop an App to help search for lost dogs.
According to My Navi, two-stage recognition is used to achieve high-precision matching. First, the deep neural network determines which dog the phone’s nose print image is close to. Then the feature is calculated by the comparison image with high possibility and the image to be identified, and the consistency is judged automatically by the feature parameters.
In order to locate the nose, the facial position detection model was taught to label photos of dogs with six points including nose, eyes and ears. This will automatically focus on your nose and crop out a square image centered around it.
Recruit Tech has teamed up with Mode Tadada, Pet Tech company SYRUP, and IBM Japan to help identify dog breeds more accurately. It is expected to use the technology to create an App that facilitates searching for lost dogs.
The article is written by: Getting your pet done http://hocacorp.com.tw