Ornithologists and app developers have flocked together to create an app that applies facial-recognition principles to birds. Take a picture of a winged creature and Merlin Bird ID will tell you its species, ecoregion, and other fun facts.
The app, which is currently limited to the 400 most common North American bird species, is the product of a three-way collaboration between Cornell Tech, Caltech, and the Cornell Lab of Ornithology. It’s easy to dismiss the concept with a laugh, but the researchers have trained the app to perform a surprisingly difficult task.
All it takes is one good science-fiction movie to overemphasize the simplicity of computers recognizing human faces, fingerprints, or retinal scans. When it comes to identifying birds, the process is harder still, due to unique challenges that computers have yet to fully overcome.
“Asking computers to identify bird species is a challenge not only because some species look so alike, but also because their shape varies from moment to moment,” Cornell ornithologists told Wired. “On top of that, photographs of birds often include complex backgrounds, and the birds may be far away or blurry,”
The team is using machine-learning technology to train the algorithm that powers the app. The more data the app collects, the more accurate it will be over time. As the system detects more patterns, it adjusts its algorithm to function better.
Despite its seemingly whimsical nature, avian facial recognition represents a toe dipped into the water of advanced artificial-intelligence technology.