Animals on camera

New reports emerge weekly about controversial uses of facial recognition technology by schools, commercial establishments, sports arenas, and law enforcement. But less known and perhaps less contentious applications of the technology are occurring in the animal kingdom.

Scientists, researchers, and farmers are using facial recognition to identify, track, feed, locate, and even diagnose disease in wild and domesticated animals. The menagerie includes pigs, cows, bears, and fish.

Like humans, animals have individually unique characteristics. A computer program can map the features of a particular animal and then later identify or track it.

The Guardian recently reported on farmers in Guangxi, China, who use facial recognition on large farms to alert them of animals sick or in distress. The software identifies the differences between individual pigs by “analyzing their snouts, ears, and eyes.” With facial recognition providing the ability to “continuously monitor, identify, and even feed their herds,” farmers have seen decreases in costs and breeding times. With facial recognition and surveillance cameras connected to a feeding system, pigs can have an individualized feeding plan that decreases the amount of wasted food.

Cargill, the Minneapolis-based, global food company, has invested in a machine-learning Irish startup, Cainthus, to offer facial recognition to dairy farmers: The software identifies individual cows and tracks their feeding habits, delivering data to help farmers make better decisions that affect milk production, reproduction, and animal health.

Another application of facial recognition in the food supply chain deals with disease in the salmon population. Bloomberg Businessweek in 2018 described how a Norwegian fish-farming company, Cermaq Group AS, planned to use BioSort’s facial recognition software to scan fish for lice or skin ulcers. Fish with abnormalities would be quarantined and treated, protecting the health of the remaining fish. A sea lice epidemic in salmon farms can cost the industry more than $1 billion in a year. According to BioSort CEO Geir Stang Hauge, early disease detection could reduce salmon mortality by 50 percent to 75 percent—a potential game changer for countries like Chile and Norway.

Moving from farms to wilderness: Distinguishing one grizzly from another barely matters to the average camper, but to University of Victoria bear biologist Melanie Clapham, identifying and tracking individual bears over the course of many years is important to her research and to the conservation of the species. In 2017 Clapham collaborated with software developers Ed Miller and Mary Nguyen to develop BearID, a facial recognition program used to catalog and track grizzlies in Canada and Alaska. In a paper published last November, the researchers noted that BearID, using a database of 4,674 bear images, correctly identified an individual bear 84 percent of the time.

Finally, closer to home, the pet facial recognition app Finding Rover helps reunite Fido and Fluffy with their owners. Multiple animal shelters across the United States, along with pet owners, upload pictures to the app’s database. ABC’s Good Morning America reported that an owner can upload a photo of a lost pet to the app and it “will scan a database of more than a million rescued or found animals that could be a match.” The app’s founder, John Polimeno, told the show that Finding Rover has helped reunite over 15,000 pets with their owners.