"They all look alike to me" is no longer an excuse when studying penguins. The same facial recognition software that helps Homeland Security identify terrorists could one day be used to identify individual penguins and monitor their populations, thereby aiding in their conservation.
Traditionally, tracking individual penguins—which is important for monitoring population dynamics, understanding migratory patterns, and assessing the health of a species—has required attaching transmitters to their backs or metal bands to flippers or legs. But transmitters are expensive, and evidence has shown that ID bands can sometimes interfere with swimming and food gathering or even injure the birds if the bands are damaged.
A team of researchers from the University of Bristol in England and the University of Cape Town in South Africa decided that current tagging methods weren’t good enough, and wondered if they could come up with a more effective, faster and less invasive method of tracking and monitoring penguins.
The result: African penguins (Spheniscus demersus), a threatened species. Similar to facial recognition software, the researchers’ system identifies unique markings on a penguin’s chest. The researchers found that once their software identified a penguin, it could re-identify that same bird 96.7 percent of the time.
Their research, conducted on rocky Robben Island off the South African coast, was published online March 25 in the journal Endangered Species Research.
The authors acknowledge that this technology is still in its infancy. The system only managed to capture recognizable patterns on 13 percent of the penguins it examined—but still, that’s a fair sight better than the naked human eye could do.
Image: Capturing penguin characteristics, from the open-access paper, "Spotting the difference: towards fully-automated population monitoring of African penguins". © 2010 Inter-Research