Backyard birders have another tool in their arsenal, with a newly updated smartphone app that can recognize a bird species by analyzing a photo.
Much like facial recognition systems, the app – The Merlin Bird Photo ID, created by researchers from Caltech and Cornell University – only needs a clear, quality photo to begin matching the picture against a database of 650 North American bird species it currently stores.
Users simply show the app their bird photo, tell the program where and in what time of year the winged wonder was spotted, and then wait a moment to see if Merlin comes up with an answer.
If it thinks it's identified the bird, the app supplies one or more possible suspects, with details about the bird's markings, behavior and breeding. Also supplied are juvenile and adult photos, sample calls from the bird, and a map showing the animal's year-round, breeding, migrating and non-breeding ranges throughout North America.
The researchers say, in a press release, that the program will be correct about 90 percent of the time, assuming a quality picture and, of course, that the bird is present in the database.
Does it work as advertised? This author supplied the app with a picture of a sharp-skinned hawk taken in his backyard and the program indeed returned the correct hawk species as a potential identification, along with a second possibility, a Cooper's hawk, which looks a lot like a sharp-skinned. The picture was somewhat dark, originally taken by smartphone at "web photo" image quality, so at least in that case the app seemed pretty agile.
Caltech engineering professor Petro Perona said the app is part of a wider vision centered around machine learning and visual classification.
"Ultimately," Perona said, "we want to create an open platform that any community can use to make a visual classification tool for butterflies, frogs, plants, or whatever they need."
