In order to observe the wildlife of the Serengeti, scientists set up hundreds of motion and heat sensitive "camera traps" around an expanse of the Serengeti National Park. On Tuesday, the team revealed the fruits of their work: a searchable database that contains more than 300,000 images of wildlife, including 40 different kinds of mammals.
To get there, University of Oxford postdoc Alexandra Swanson and her team had the help of about 28,000 volunteers, who sifted through a total of 1.2 million images collected by the cameras between 2010 and 2013 to identify the wildlife they captured. A paper on their process and findings was published in Scientific Data, one of Nature's online publications.
Many cameras captured bursts of three images at a time, which means that you can also see a little bit of movement if you play the images in sequence. Instant animal gifs!
There's even a category of "human" photos taken of researchers and park workers during installation and periodic camera inspections.
Volunteers have also added tags to the images beyond just species identification. The "selfies" tag appears to contain a lot of images of animals looking directly into the camera, like this Grant's gazelle:
At least one intrepid volunteer went through and tagged together a sequence of separate night images that effectively document a lion killing and eating a zebra.
Some cameras caught rarely-seen creatures, like this zorilla:
And of course, there's a series of tagged "young" animals, from which a disproportionate number of images in this post were drawn, because that is what this reporter's heart says is the right thing to do.
"Without volunteer help, the research wouldn't have been possible," Swanson said in a statement.
"We all know that people are good at pattern recognition, so harnessing the power of volunteers will become increasingly important for ecology studies," Swanson said. "We can engage people with no scientific background to help in producing publishable scientific research at a scope and scale that would otherwise have been impossible."
The results are viewable at Snapshot Serengeti, where Swanson and her team uploaded the images for the volunteers to sort. The database was designed by Zooniverse, which hosts tons of these kind of citizen-powered scientific projects.
You can still try it out, if you're so inclined. The system isn't flashy, and it's hard work, but the database gives you all the information you need to distinguish a hartebeest from a wildebeest.
But unless you're itching to make tough calls on horn shapes and fur patterns, looking at one of Zooniverse's newer projects might be a better use of your time. According to the paper, the classifications made using Snapshot Serengeti are already around 97 percent accurate.