Until now, birdwatchers relied on cumbersome field guides to identify the species they observed in their backyards. But as nature-lovers embrace the digital age, birding manuals are getting a makeover complete with cutting-edge facial recognition software.
Researchers from Columbia University and the University of Maryland released a free iPhone app last month called "Birdsnap" aimed at helping amateur birdwatchers identify species. Professor Peter Belhumeur of Columbia, who supervised the project, says his inspiration stemmed partly from his admiration for birdwatchers' ability to accurately identify species on sight. "For the rest of us, it's a much harder process," he explains. "So we made this app to bridge that, to help people without training to get a leg up on identification."
On its surface, the program seems simple enough: You snap a photo of the tricky bird in question and upload it to the program, which processes your photo and reports back with a list of possible matches. In reality, however, a complex facial recognition software is at work, identifying the various parts of the bird—feet, neck, wings, tail—and comparing your photo to a database containing thousands of images. The program uses characteristic bird markings—for example, the distinctive white "eyebrows" on a Carolina wren—to come up with the best possible match.
The app is not without its quirks. Facial recognition software works best with clear, high-quality images—hard to come by when snapping photos of birds on the go. I uploaded a photo I took in New York City of the familiar city pigeon, and the app told me I had just seen the rough-legged hawk—a species usually found in northern Eurasia this time of year. The dark-eyed junco was its second guess, and the correct answer rolled in at number three.
On the other hand, the program immediately matched a photo of a female house sparrow. Belhumeur says his team plans on adding a feature to the app soon that will allow users to zoom in on their bird photos after taking them, potentially increasing the app's effectiveness.
Photo identification is just one of the app's features. Birders who are interested in improving their skills or brushing up on their avian knowledge can browse Birdsnap's digital field guide, which features 500 common North American species accompanied by photos, short descriptions, range maps, and lists of similar species with ways to tell them apart. The app also keeps track of migrating birds' movements around the country, and has a complementary website with similar features as the app.
The app is one in a line of feature recognition programs Belhumeur and his team have developed: Leafsnap, released in 2011 after 10 years of development, and Dogsnap, released in 2012. The developers may not be stopping there. Belhumeur says he hopes to continue adding features to Birdsnap, such as the ability to identify species by their calls. He also hopes to create a set of "smart" binoculars in the future, which will use similar technology to identify birds in real time.
For now, Birdsnap lets nature-lovers swap field guides for smartphones, allowing for easy, on-the-go birding. If anything, it's one more way to differentiate between all those pesky sparrows—and it may be the first of a new arsenal of tools for the tech-savvy birder.