Facial Recognition...For Primates
Facial recognition is becoming more commonplace in our society, from unlocking our phones to catching criminals. But now this technology is being used on other species of primates too, and is one of those great examples of innovative techniques being adopted by conservation that makes me feel hopeful for the future!
More than 60% of primates are endangered. They are one of the most endangered groups of mammals - our cousins. I think that’s pretty shameful, and I’m not alone. Some clever conservationists are trying to do something about this statistic, and they have created a clever solution.
A traditional way of conserving animals is to study them in the field, monitor their lives and population levels over time. This involves capturing the animals and taking measurements or samples, and sometimes fitting tags or tracking devices. Despite the best efforts of conservationists to minimise the distress this causes animals, it can adversely affect them. It may become agitated or injured, or in the worst scenario, cause death. For highly social animals like primates, it can disrupt their social behaviour, and for worn out conservationists, it can be time-consuming. And yes, if you want to be cold-hearted about it, it costs money, in veterinary services, anaesthesia and tracking devices.
In 2017, it was announced that a team of lemur biologists and computer scientists from various universities in North America had created a facial recognition system for lemurs called LemurFaceID. The new technology, modified from human facial recognition methods, can identify individual lemurs for the first time, ever! And it can do so highly effectively - with an impressive 98.7% accuracy if it has two images of the individual’s face.
Previously, all that could be used to identify lemurs is unique markings (eg scars or injuries) or differences in body size or shape, all of which may change over time. Long-term study of individuals was therefore problematic.
But LemurFaceID will make this much easier, allowing for information to be collected on frequency of reproduction, how long they live for, mortality rates, and ultimately whether the population is growing or in decline. Compared with traditional methods, it will be more accurate, quicker, more cost-effective and crucially, non-invasive. The lemurs will be left alone to be lemurs, without human interference.
It could also be a valuable tool in the fight against illegal wildlife trafficking - a captured primate could be photographed and identified to determine its origin, which can give vital clues about its capture and help improve efforts to prevent future crimes.
The species that was used in the study and creation of LemurFaceID was the red-bellied lemurs. Like many lemurs (endemic to Madagascar), this species faces threats from habitat loss, as large tracts of land have been cleared for illegal logging and agriculture. Another primate facing a similar threat is the golden monkey from Africa, which has lost so much habitat that they are only found in a handful of national parks. This species was the focus of a 2018 study which documented the creation of a new and improved primate facial recognition system.
This one is called PrimNet, and claims to be even more accurate than LemurFaceID. To recognise individuals, it uses “convolutional neural networks” - no idea either, but apparently it is technology that is inspired by artificial intelligence and used in robots and self-driving cars. And the best thing about this one is that it comes with an app! PrimID enables field researchers to simply snap a photo of a golden monkey, upload it to the app and said monkey will be identified.
It is hoped that this technology will be expanded to cover many other species, both of primates and of potentially a whole host of other animals that have variable facial hair and skin patterns, such as bears, sloths and raccoons. This is definitely a case of tech for good!