You Can Now Identify Birds With Just A Photo
A stroll in the park can help ease your mind or even help you boost your math skills. You can be sure to encounter some usual creatures like squirrels, rabbits and pigeons, but also some unusual avian creatures. Figuring out the names of these birds is a tough task. If only there were an app to find out what these uniquely stunning and beautiful creatures are called...
That's where Merlin Bird ID comes in to save the day. It's a program that lets you upload the pictures you take of birds, and then uses computer-vision technology to present possible species. The mobile app, developed by researchers from Cornell Tech and the California Institute of Technology alongside the Cornell Lab of Ornithology, asks the user a series of questions to generate a list of possible species. These possibilities are even accompanied by bird calls.
Jessie Barry, a researcher at the ornithology lab, has said that not only is the tool "a pretty big deal" in the world of birding, but also:
"Ten years ago, people were thinking, 'Can you imagine if binoculars could take pictures and identify birds?' And the fact that we can actually snap an image and have the computer say what it is, that's something you only would have dreamed of."
The app asks five simple questions like "what colour was the bird?" and "what size was the bird?" to help narrow the searches. Merlin is a way for beginning and intermediate bird watchers, or even those just a little bit curious about birds, to learn about (specifically) North America's most common birds.
Serge Belongie, a computer science professor at Cornell Tech, said:
"We're not aspiring to remove the human element and experts from the loop. What we want to do is use their time much more efficiently."
The engineers have used a vast array of resources available to them: those associated with the ornithology lab, for instance, have helped to develop and constantly refine the data; Merlin is also making use of the bird sighting information from eBird.org.
Merlin Bird ID is an extention of the Visipedia research project, a joint research project between Pietro Perona's researchers group at Caltech and Serge Belognie's researchers group at Cornell Tech. The research project was started by engineers about five years ago as a "visual Wikipedia". In particular, the program can identify 400 of the most common North American bird species.
The current research will be presented today at the Computer Vision and Pattern Recognition conference. The researchers plan to expand on their database to include other geographic regions and even more bird species. They are inviting us to use and engage in the databuilding process. The more people use Merlin, the more accurate it becomes.
No comments:
Post a Comment