Communications of the ACM
New Website Can Identify Birds Using Photos
The Merlin Bird Photo ID is a program that can recognize 400 of the most commonly seen birds in the United States and Canada.
Credit: Cornell University
Cornell University researchers have developed the Merlin Bird Photo ID, a bird photo identifier that can recognize 400 of the most commonly seen birds in the United States and Canada. The program is able to correctly identify the bird in the picture among the top three results about 90 percent of the time, according to Cornell researcher Jessie Barry.
In addition, the website is designed to keep improving as more people use it. Users upload an image of a bird and tell Merlin where and when the picture was taken. Users also must draw a box around the bird and click on its bill, eye, and tail, which helps orient the computer vision system. Merlin uses artificial intelligence techniques combined with millions of data points submitted by users to determine the most likely species, including photos and sounds.
"Computers can process images much more efficiently than humans--they can organize, index, and match vast constellations of visual information such as the colors of the feathers and shapes of the bill," says Cornell professor Serge Belongie.
The system also relies on more than 70 million sightings recorded by bird watchers in the eBird.org database.
From Cornell University
View Full Article
Abstracts Copyright © 2015 Information Inc., Bethesda, Maryland, USA
No entries found