However you may feel about facial recognition technology, it’s at least saving lives—of the seals, anyway. Years down the road, some pups will grow up in a more abundant environments because scientists will have better understood their behavior and how to conserve their lives.
Researchers at Colgate University in New York have developed ‘SealNet’, a database of the faces of 400 harbor seals residing at Casco Bay in Maine. The tool aims to be a non-invasive way of tracking seals and having a closer look at their ecosystem.
The findings, published in the Ecology and Evolution scientific journal in April, details that the system has been able to identify the seals with a nearly 100% accuracy.
Biologists have long traced the footprints—or fins—of marine mammals using satellite technology. Artificial intelligence propels these studies into the future, Jason Holmberg, the executive director of machine-learning conservation research firm Wild Me, tells the Associated Press. His company is in talks with SealNet’s creators to potentially collaborate.
SealNet is able to distinguish seals’ facial features in photos, then observe those unique markers such as eyes and nose shapes in the wild. Previously, biologists utilized a similar system for primates, called ‘PrimNet’, on seals, but the newer and more specialized SealNet has turned out to be more powerful.
As per Krista Ingram, a biology professor at Colgate University and one of the researchers for SealNet, storing pictures of seal faces and harnessing machine learning to track the mammals will help scientists understand their patterns more clearly, since the mobile creatures would be hard to catch up with in open waters. The hope is to devise ways to better protect the pinnipeds.
In spite of its accuracy, the tool still makes mistakes sometimes. As per Hakai Magazine, there have been a few instances of SealNet incorrectly detecting seal faces in body parts, vegetation, and rocks—even the human researchers agree that the marks in some rocks look like seal faces, and they had to do a double take. This means the software’s results may still require subsequent human assessment.
Harbor seals were viewed as pests in 19th-century and early–20th–century New England, and they were hunted down fervently because the authorities were doling out bounties for their extermination. Thankfully, efforts to rebuild their population have since proven to be successful, the Associated Press reports.
However, other species aren’t thriving as well, including the Mediterranean monk seals, of which only a few hundred seals are known to be alive and are said to be the most endangered in the world.
The team is now looking to fine-tune automation for SealNet so as to open its resources to the wider community. Climate-loving citizen scientists could eventually help out and record new seal faces, too. In time, the seal face-finding tool may not just extend to observing the Mediterranean monk seal but also other rare families like the Hawaiian monk seal, and even whales and dolphins.