Latest in Science

Image credit:

Cornell's website can ID bird species through photos


Casual birdwatchers may want to bookmark Merlin Bird Photo ID, a website created by Cornell University and the Visipedia research project. Thanks to powerful artificial intelligence techniques, the website can identify birds in photos you upload, so long as they're species found in the US and Canada. You will need to draw a box around the feathered creature, click its beak, eye and tail, as well as identify where and when you took the picture. But after that, the system will do the rest: it will sift through the numerous photos taken by other birdwatchers, including the 70 million images in the database, to find ones that match your image.

It will then present you a list of the best matches "within seconds," along with their photos and recordings of the sounds and songs they make. Like any other system that uses machine-learning tech, the more people use it, the better it becomes at identifying different species. As Cornell Tech Computer Science professor Serge Belongie said:

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. The state-of-the-art in computer vision is rapidly approaching that of human perception, and with a little help from the user, we can close the remaining gap and deliver a surprisingly accurate solution.

The website is open to anyone and free to use, but it doesn't work on phones and tablets yet. However, its creators promise that when it does become good enough to ID birds in smartphone photos, they'll add the feature to Cornell's Merlin app for Android and iOS. That application helps you identify avian species, as well, but it's by asking five questions about the bird you spotted.

From around the web

ear iconeye icontext filevr