top of page

AI Case Study

LemurFaceID identifies an individual lemur with 98.7% accuracy using facial recognition technology

LemurFaceID is an app that uses facial recognition technology to research lemur population and study the conservation of the species. The software, developed by researchers at Michigan State University, The George Washington University, City University of New York, and the University of Arizona has been trained on photos to identify lemurs using facial recognition technology. It is claimed that the system can identify an individual animal with 98.75 accuracy.


Project Overview

"Facial recognition software, by contrast, can identify an individual lemur almost immediately, and all that’s required is a photo. Tecot was part of a team of two other lemur researchers and four computer scientists, from several universities, who developed the software. While facial recognition for humans focuses on the geometry of facial features, such as the distance between eyes, the system for lemurs looks for unique patterns in the animal’s fur.

For lemurs, the new software will make it possible to better understand their lives. “For the answers to really interesting evolutionary questions, or questions about population dynamics that are important for conservation efforts, I think it’s really a requirement that this is done,” says Tecot. “We couldn’t do it otherwise.”

Along with the majority of other lemur species in Madagascar, red-bellied lemurs are listed as “threatened”–at risk from illegal logging and mining as well as the illegal pet trade. Facial recognition software could help law enforcement identify an animal sold as a pet; it could also help conservation experts plan actions, and help increase support for conservation in general.

As people document their visits to Madagascar’s Ranomafana National Park, where the researchers work, their photographs could help grow the researchers’ database and improve the facial recognition software. And, in turn, the software could connect visitors to the animals by telling them the name of the lemur they just put on Instagram.

“We think one of the things that can improve conservation is feeling more connected with the park and with the animals–that’s true of tourists, guides, and locals, anyone who goes into the park–so there’s an emotional connection,” says Tecot. “People can go into the park, snap a photo . . . that can create a more personal experience so people know who they’re taking a picture of and they get a little bit of information about them”."

Reported Results

"LemurFaceID, new software developed for lemur research, can identify an individual lemur with 98.7% accuracy, making it easier to track a population over a long time and plan for conservation."




"Until now, researchers studying red-bellied lemurs in a rainforest in Madagascar have relied on their own ability to recognize a particular lemur. “Some have individual scars; some have eyebrows that look like they’re scowling,” says Stacey Tecot, an anthropologist from the University of Arizona who has spent the last 17 years studying lemurs. “But visibility in the forest is pretty difficult.”

It takes time to train new researchers to identify individual lemurs in the field, and it’s difficult to keep track of those individuals over time. (It’s not known how long red-bellied lemurs live, but they may live into their twenties or thirties.) It’s also difficult to track populations over large areas.

The other alternative–capturing animals and adding tags or tracking collars–can be stressful and expensive. Tags and collars also potentially injure wildlife, especially tree dwellers like lemurs. For species that are already threatened or endangered, it’s one more threat."



images and photographs

bottom of page