How Do You Pick a Lemur Out of a Lineup? This Software Makes the Leap
Facial recognition software can identify individuals, helping researchers conserve the endangered primate
/https://tf-cmsv2-smithsonianmag-media.s3.amazonaws.com/filer/19/f9/19f94d18-1422-445f-906d-d7dd80685546/eulemur_rubriventer_001.jpeg)
It sounds like something out of a crime show: police pouring through data from surveillance cameras, using facial recognition software to nab the perp. But now, researchers have adapted this software for use in the forests of Madagascar, identifying and tracking the whereabouts of endangered lemurs.
As the BBC reports, the software, known as LemurFaceID, allows scientists to more effectively track and protect the primates. The software can distinguish individual lemurs from digital photographs with greater than 97-percent accuracy. Researchers hope the tool will improve conservation of the species while providing a more humane, noninvasive way to identify individual lemurs. The team recently published their work in the journal BioMed Central Zoology.
To track lemurs, scientists traditionally trapped and tagged individual animals. They cataloged their physical characteristics—body size, markings, notable scars or injuries. But tracking these lemurs as their appearance changes over time is both time consuming and challenging, hampering long-term studies.
“[We] weren’t particularly satisfied with the common approaches used in lemur research,” Rachel Jacobs, a co-author on the paper, tells the BBC. “[S]o we aimed to do something different with red-bellied lemurs, and we sought the expertise of our computer science collaborators.”
To develop the software, Jacobs, a biological anthropologist from George Washington University, turned Anil Jain, a biometrics expert and distinguished professor at Michigan State University.
Jain and his students in the computer science department created a dataset comprising 462 images of 80 red-bellied lemurs primarily taken in Ranomafana National Park in Madagascar. The researchers also included an additional 190 images of other lemur species to help expand the software’s capacity. To identify an individual, LemurFaceID first identifies its eyes and then analyzes the characteristics of each surrounding pixel in the image.
“Like humans, lemurs have unique facial characteristics that can be recognized by this system,” Jain tells MSU Today.
The new software will give lemur researchers and conservationists a new tool for tracking lemurs over time. Long-term data provides researchers with crucial metrics to measure population growth and decline, like the rates of infant and juvenile mortality.
The software could also aid in the fight against illegal captures of the big-eyed primates. With only a clear digital image, locals and tourists can report sightings to law enforcement and researchers to quickly identify captive lemurs.
The researchers believe LemurFaceID could be adapted to protect other mammals with variable facial and skin patterns as well. Jain tells MSU Today that he believes the software could work for bears, red pandas, raccoons and sloths.