Tags and chips are poor pet ID tools. Not all pet owners like microchipping since tags readily come off. Chip breakage and obsolete ID databases plague even the most comfortable microchip users.
The challenge motivated Jesse Joonho Lim and Ken Daehyun Pak to create Petnow, a face-scanning app that can recognise cats and dogs. Petnow, which has received $5.25 million from Daedeok Venture Partners and DigiCap at a $24 million valuation, is in the Startup Battlefield 200 at TC Disrupt 2023.
Lim co-led Chips&Media, a semiconductor firm that went public, before starting Petnow in 2018. Pak, like Lim, holds a PhD in electrical engineering and worked as an AI video processing researcher for almost a decade before joining Petnow.
Petnow uses a camera-based scan of a pet’s face from an Android and iOS app. Petnow builds a pet’s unique biometric profile using AI based on 200,000 photos of dog and cat snouts from the Petnow team and users’ pets.
Petnow says its algorithm intelligently detects and crops out dogs and cats, preventing the app from mistakenly capturing a family member behind a pet.
For dogs, Petnow records a “nose print.” Yes, noseprint. The business argues that a dog’s nose is as distinctive as a fingerprint and doesn’t alter, giving it a dependable way to recognise pups. Petnow examines cats’ “facial contour,” which is unique due to their “grooming habits.” (That sounds more suspicious to this reporter than a dog’s nose print, but I digress.)
Lim and Pak envision Petnow being used to register pets without vet visits, detect lost pets, and develop “pet IDs” for insurance checks.
“The pet identification market may not be mature at the moment, but it will eventually become big,” . Pet identification technology is a core product that consumers can use long-term, unlike viral items. The market is promising since dogs should have IDs like people and their data should be stored up to create an ultimate pet platform.
Does the tech operate as advertised?
Petnow promises “99% accurate” cat and dog identification algorithms. It’s commonly known that even the finest image-analyzing AI has bias, intentional or not.
Six Black persons have been wrongfully arrested by police using facial recognition technology. Facial recognition systems are often trained on datasets without enough Black faces, producing prejudice. Or, they’re trained on mugshot datasets with a lot of Black faces, many photographed in poor, grainy lighting, making it hard for the computer to differentiate faces.
Even specialists have trouble distinguishing breeds, let alone creatures of the same breed. A recent survey of 5,000 dog specialists discovered that only a small proportion could identify even one breed from their DNA.
Petnow boasts that its training database is increasing and that AI ensures pet photos are bright and sharp. Petnow says it doesn’t share user or pet data with third parties without agreement and offers an option to remove data at any time, apparently anticipating data privacy concerns. Petnow cites an IEEE Access research co-authored by its data scientists that indicates its dog nose-print detecting tech was over 99% accurate.
The work dates to 2021, when the training dataset was likely smaller. Petnow claims to be working on a companion paper for their cat face-recognizing algorithm, but it hasn’t released it.
The stakes are high. An computational error might impede a family’s hunt for a lost pet or cause a vet to draw up the wrong animal’s immunisation data.
Given Petnow’s delayed adoption by pet care providers and shelters, those aren’t imminent concerns. Despite having 70,000 members, Petnow has only recruited five unidentified commercial and public sector customers in France, Toronto, and South Korea.
Petnow burns $150,000 every month pre-revenue. However, it expects a contract with Korean local and international pet insurers by October and pet registry pilots in France and Canada.
Due to the pandemic, pet ownership has increased and people spend more time with them. Lim and Pak claimed there’s significant growth potential. South Korean government pet registration and pet insurance affiliation programmes are on the way, and we’ll make our solution enterprise-ready for North America and Europe soon.”
I just hope Petnow and its competitors give proper pet identification. Failure would be irresponsible; deceiving pet owners seems like cruel false advertising.