Automatic photo ID—thanks to your eBirding

What's that bird? Merlin says Osprey!

Bird identification can be a challenge, especially for beginning birders. While solving these challenges can be a compelling part of the fun of birding, at times it can also be frustrating. The Cornell Lab is interested in building tools to help people become better birders, and also help engage new communities around the world in the joy of birding. 

Imagine if everyone who carried a smartphone or digital camera was one shutter click away from identifying a bird? For 650 species, this is now a reality. Last week, the Merlin team released Photo ID—a new feature in the free Merlin app that provides real-time, offline, bird identification. Of course, you should still double-check the results from the app, but we’ve found the computer to be unnervingly accurate! Download the app and see if you can stump it. The most exciting part of this is that you make it possible, thanks to your sightings, photos, and eBirding. Want to expand Merlin to more than 650 species? So do we—but we need your help! Click the full article to see how you can bring Merlin to your backyard.

The automated photo identification is powered by computer vision models in Merlin, developed in partnership with the Visipedia team. Here’s when you come in: the wellspring of Merlin’s power is your eBird checklists.

When you give Merlin a photo to ID, you’ll notice that it asks where and when it was taken. The reason for this is your eBird sightings: Merlin looks at the frequency of eBird reports from the area, and is able to give results more accurate than any field guide by using the local information on bird distribution provided by eBirders. And the computer vision tools that are used to identify the image? They’re developed from the images that you’ve added to the Macaulay Library—more than 1.85 million photos over the past 13 months. Once we have the images, volunteer annotators have used the MerlinVision tool to draw boxes around hundreds of thousands of photos, giving valuable clues to the computer vision models. If all of this sounds exceedingly complicated, no worries. The moral of the story is that what you contribute does make a real difference. We couldn’t do this without you.

Anywhere you go (even offline), Merlin can identify what you’re seeing.

Anywhere you go (even offline), Merlin can identify what you’re seeing.

The exciting news? 650 species is just the beginning! In 2017 we’ll be expanding Merlin southward through Mexico and then much of Central America, with the potential to grow only limited by the available information. Every time you go out in the field, you have a chance to help bring Merlin’s powerful tools to your backyard. Do you live in Australia and have a flock of Galah at your local patch? Take a photo every time you see them. How about a Common Chaffinch in your backyard in Germany? We’d love your daily photos at the feeder. Whether it is a Jungle Babbler in India, a Bananaquit in Brazil, or a Hadada Ibis in South Africa, your photos provide the building blocks to a global Merlin Photo ID!

How to bring Merlin to your backyard

Submit eBird checklists—This provides the required information on where birds occur in your area, allowing Merlin to be as accurate as possible.

Add your bird photos to eBird checklists—Each photo is an essential piece of information: once we have 1,000 of a species we can create these amazingly accurate photo identification resources. Check out Merlin’s Most Wanted photos here.

Look at pictures of birds with MerlinVision—A perfect excuse to page through many lovely bird pictures, all you have to do is draw a quick box around each photo.