To ID birds, try facial recognition

SAY CHEESE  The Birdsnap program maps birds’ bodies to identify species, such as a bohemian waxwing shown, in photos taken from different angles.

themadbirdlady/flickr (CC BY-NC-ND 2.0)

Birding just got easier. No need to page through guidebooks looking for the mystery bird you just spotted — all you need is an iPhone and the new Birdsnap app, or a digital camera and computer. Snap a picture in the app or upload an image to the Birdsnap website, click on the bird’s eye and tail, then enter the location and date where the bird was found. Sophisticated algorithms detect parts of the bird, such as the beak and the belly, then maps the bird according to “a common coordinate system from which [Birdsnap] can extract features for the identification system,” explains one of Birdsnap’s creators, computer scientist Peter Belhumeur of Columbia University.

This is similar to how facial recognition systems map features such as the nose and eyebrows to identify people, even when they’re not facing a camera head-on. After a few seconds, Birdsnap gives the user its best guess of the bird’s species. Birdsnap currently lists only North American species, and it isn’t perfect in its identifications — Belhumeur calls it a “work in progress” — but it can help narrow down potential species. Then even an amateur birder can use Birdsnap’s descriptions and audio of bird calls to make the ID. Even better, Belhumeur notes, the system teaches you how to recognize birds from key parts so that later, you won’t even need your automatic field guide. 

Identify birds using your photos at birdsnap.com.

Sarah Zielinski is the Editor, Print at Science News Explores. She has a B.A. in biology from Cornell University and an M.A. in journalism from New York University. She writes about ecology, plants and animals.

More Stories from Science News on Animals

From the Nature Index

Paid Content