Russian security products supplier Goznak will use artificial intelligence to track missing pets. They will find your beloved four-legged friend based on its photo using biometrics and artificial intelligence. What other opportunities can this technology offer besides finding a lost pet?
How can technology help find missing pets?
According to Mars Petcare, the population of pets in Russia is as big as 63.5 mio, with dogs and cats accounting for 90% of all pets in Russian families. Stray animals have been a problem in Russia for a long time. Dogs and cats can find themselves on the streets not only as a result of human irresponsibility. The sad fact is that pets often run away and get lost. Increasingly, people are starting to search for missing animals through social media and various pet search sites.
Attempts to use technology to identify dogs have been made for a long time. Canada, China and the United States have the best-known dog biometrics projects. Identification of dogs by nose prints has become very popular. The shape and print of a dog’s nose is as unique to that dog as a human’s fingerprint.
Goznak also developed a project to identify dogs by nose, but scrapped it later — the solution was inconvenient because it required a high-quality photo of the pet’s nose, with a good camera, perfect lighting, and the dog standing very still while the photo is taken at close range.
Instead, Goznak proposed a technology for identifying pets using a photo of their face, which will help to quickly find your lost four-legged friend. Unlike nose print identification, which requires a photo of the dog’s nose taken from no more than several centimeters and with maximum clarity, this technology is less demanding when it comes to shooting conditions. Images made from 1.5 to 2 meters away can be successfully used for identification. The technology also does not require any tailored infrastructure or expensive equipment.
Why Goznak develops pet identification technology
Goznak is actively developing its digital division. Having proprietary biometric identification solutions, the company participates in the Integrated Biometric System (IBS) that makes it possible for bank customers to make requests remotely and receive services without visiting a branch. Goznak has a history of cooperating with the Russian Cynological Federation that has shown interest in biometric identification of dogs. The first technology they tested was dog nose identification. Then our team developed a solution for identifying dogs based on the images of their heads. It is important to note that the solution was developed using Goznak resources and funded by its own budget.
We also conducted a thorough market analysis, primarily in-depth interviews with pet owners, breeders and vets, to generate a pool of hypotheses about the potential product. Those hypotheses were verified using quantitative methods, or by surveying over 1,000 dog owners. The outcome indicated that as a business niche, pet identification is still available and we can enter it.
Our goal is to make pet identification by photo an alternative to existing identification technologies such as implant with unique 15-digit codes that stay with animals forever. Few owners are willing to have their pets tagged and many believe this identification method is inhumane.
We are also seeking opportunities to make it a commercial service. The pet goods market is enormous so building a platform for pet owners will bring substantial profits. We have researched around six options for monetizing this solution and are currently working on a partnership agreement.
How it works
The dog face identification technology is based on artificial intelligence. An image is fed to a neural network that determines whether there is a dog in the image. The network will then identify the face among other objects in the photo and map out the pet’s detailed head shape. If the angle meets requirements, the dog face image is used to build a vector of unique features and find their match in a lost and found pet database.
Goznak plans to launch a lost and found pet website to test and demonstrate the technology in real settings. We will have a message board for two types of messages, lost pets and found pets. Currently, owners have to scroll through tons of messages to find their pet.
Our AI will automatically scan the entire database of images and show matches. The system’s ability to automatically match lost and found pet images is what makes it stand out. More than that, we want charities and volunteers to use our website as yet another way to find lost pets a new home.
We have joined efforts with our partners to have pet identification by face photo recognized as an official method. Once we have extensive public support, the pet database will be frequently updated, which should increase the probability of finding one’s lost pet quickly. Thus, it will help to promote responsible pet ownership and make it more humane. Our special project devoted to search for lost pets will develop the volunteer movement and charitable activities.
In the future, photo identification of pets will be used to search for similar-looking dogs and cats. For instance, you saw a puppy you liked and the technology will allow you to find a similar looking one in a nearby kennel.
By Georgy Kornilov, Director of Goznak Research Institute