Zoe Sagalow, special to WTOP.com
WASHINGTON -- Want to know what kind of bird is perched in the backyard? Now there's an app for that.
David Jacobs, computer science professor at the University of Maryland at College Park, worked with Peter Belhumeur, computer science professor at Columbia University, to develop an identification app called Birdsnap. Jacobs describes it as an "interactive field guide." Upload a photo of a bird to the app and he says it will use computer vision to recognize a few species that might match the bird in the photo.
The app also shows which species are similar and might be easy to confuse with each other, including what parts of the birds are similar and how to distinguish between species. The app also includes information on birds in the U.S., such as what time of year they might be found in which areas and photos.
Jacobs says he sees the app as useful to people who enjoy observing and identifying birds in their backyards.
"In general people find that traditional field guides can be hard to use because they require… a dichotomous key, and it's often not easy for someone who's not very experienced -- and even sometimes for people who are very experienced -- to make use of these," Jacobs says.
He says he also hopes the app will help collect biodiversity data. Biologists doing biodiversity research have used Leafsnap, a similar identification app Jacobs developed that helps identify species of trees in the northeastern U.S.
"There's a tremendous interest by biologists in understanding the distribution of different species of organisms on the planet and, in particular, in understanding how those distributions might be changing, due to factors like human development or climate change," Jacobs says.
Leafsnap was the first of a few nature identification apps that Jacobs developed along with Belhumeur, with whom he has been collaborating for more than 10 years.
"We'd both worked on other problems in… visual recognition, things like face recognition," Jacobs says. "We were kind of brainstorming for other applications of recognition that might be fun and might be interesting."
When they got the idea to use recognition software to identify species of plants, Jacobs says, they reached out to John Kress, who headed the Smithsonian Institution's botany department, and worked with him on Leafsnap using a grant from the National Science Foundation.
Near the end of their work, the iPhone came out, and they decided to make an iPhone app -- Leafsnap was released in 2011 and has had more than 1 million downloads, Jacobs says. A new version of Leafsnap came out earlier this summer for identifying trees in the U.K., prompted by interest in the original Leafsnap from the National History Museum of London.
"It's really fun to work with people in other disciplines," Jacobs says. "I've learned something about biology… [and] a lot about how scientists in different disciplines approach their field and what kind of questions they find interesting...
"As computer scientists, we're really interested in fundamental questions about recognition: How is it possible to recognize objects visually [and] how are people able to do this?" he added. "Species identification is a really challenging testament that forces us to come up with new fundamental ideas about recognition."
Later, Jacobs and Belhumeur developed a similar app for recognizing species of dogs, called Dogsnap, which has had about 75,000 downloads.
Jacobs says the computer science involved in developing Birdsnap "really stresses what computer vision can do because birds are complicated animals: parts move in complex ways, they vary a lot, the beaks of different birds can be very different. So there are a lot of interesting challenges for us as well as a real need for it."
© 2014 WTOP. All Rights Reserved.