There are many interesting biological questions a biology degree doesn’t necessarily equip you to answer. For instance, as a bio major at one of the world’s top ornithology research universities, I managed to skate by without learning diddly squat about birds.
Merlin can recognize 400 bird species, including the Blackburnian warbler shown here.
I know, I know: No one to blame but myself. Thankfully, Cornell’s Lab of Ornithology, in collaboration with the Visipedia research project, has just released a handy new tool to catch me and the other bird delinquents up to speed. Think facial recognition, only for finches and warblers. Merlin Bird Photo ID’s image recognition algorithms are capable of identifying 400 of the most commonly sighted birds in the US and Canada.
To get started, you simply upload an image of the mystery bird in question to the website and enter the location and date that the picture was taken (you can also leave those fields blank). You’ll be prompted to draw a box around the bird, click on its bill tip, eye, and tail tip. Merlin does the rest, querying the eBird.org citizen science database’s 70 million entries to present you with the most likely species matches, including photos and sounds.
The website is still in beta, and the developers are the process of building a mobile app. But that’s no reason you shouldn’t give it a go now: Because the photo identifier uses machine learning techniques, it has the potential to improve the more we use it. Personally, I’m excited to see more guidebook-like applications of image recognition technology in the future. Just imagine a world where you could go on a hike and, using your smartphone, quickly determine whether every plant, mushroom, spider or snake you came across was dangerous or not. Sure, it might take a little magic out of the experience, but it would also save a lot of us from winning Darwin awards.
Top image via Cornell University