Researchers at Cornell’s Ornithology Lab are working on a machine-learning bird-identification system. Called Merlin Bird Photo ID, the system uses technology similar to that used in facial recognition, looking for patterns in the images of birds to identify the species.
As it’s fed more images, the system refines its capabilities, becoming more advanced and accurate in a system of machine learning. At the moment, it’s able to identify about 400 species of birds common to North America, but that number is increasing, as is the system’s accuracy. Ultimately, the researchers would like to develop it enough that it could be integrated into a mobile app, able to correctly identify bird species even on low-quality photos; and the same technological basis could be expanded to other areas, helping to identify other animals, plants, and so on.
It’s an exciting and unusual area of biometric identification. While facial recognition technology for humans is already relatively commonplace, it’s already being used with animals in commercial applications like the Finding Rover app for the identification of missing dogs, and it could soon extend even to inanimate objects, allowing artificial intelligence systems theoretically identify everything in a given image. That’s probably a long way off, but in the near future birding enthusiasts stand to benefit greatly.