Chinese AI Firm Develops Nose Print Recognition to Find Lost Dogs
Chinese AI Firm Develops Nose Print Recognition to Find Lost Dogs

A Chinese tech company has developed a software programme that can identify dogs by their noses to help owners find their lost pets. The system, created by Beijing-based artificial intelligence start-up Megvii, uses a deep learning framework to analyse the unique markers on a dog's snout.

Similar to human fingerprints, each dog nose print is unique and remains unchanged throughout the animal's life. Owners can scan their dog's nose from multiple angles using a smartphone camera, and the AI-powered software will create a unique profile based on distinguishing markers.

The technology covers two scenarios: a 1 vs 1 comparison to confirm if it is the same dog, and a 1 vs N search to identify which nose print belongs to a specific dog. Megvii told MailOnline that the software has an accuracy rate of 95 per cent in the first scenario and took less than a year to develop.

Wide Pickt banner — collaborative shopping lists app for Telegram, phone mockup with grocery list

In addition to helping locate lost pets, the technology is cheaper and less invasive than microchip implants. It can also assist urban authorities in managing pets, curbing dog attacks, and identifying irresponsible ownership, such as walking a dog without a leash or failing to clean up after them.

Developers faced challenges in collecting good quality data because dogs rarely stay still for photos, leading to poor angles or low definition nose prints. Despite these issues, Megvii plans to expand the technology to identify other types of pets.

The number of pet owners in China has risen due to economic growth and increased disposable income. Urban areas are home to more than 91 million dogs and cats, according to a 2018 industry report. Many cities have implemented new rules on dog-keeping to improve public safety and reduce tensions between pet owners and non-owners.

Pickt after-article banner — collaborative shopping lists app with family illustration