16 animal selfies that capture life in the Serengeti
Do you think warthogs are camera shy? Are anteaters more photogenic than ostriches? Thanks to an international project documenting the mysterious life of animals in eastern Africa we might be able to find out.
Researchers set up 225 cameras across more than 400 square miles of the Serengeti region in northern Tanzania. These high tech cameras take a snapshot when any object, usually an animal, warmer than the surrounding temperature moves into view. Images are snapped in bursts of three to document animal activity, even at night. The cameras capture it all — monkey selfies, hippo rear ends, birds attempting to eat the camera and a gazelle posing for a family portrait. Occasionally a rare vehicle, butterfly or even tall blades of grass heated by the sun will trigger the camera to shoot. After weeks of collection, researchers upload the images to the website Snapshot Serengeti and ask the general public for a hand in data analysis.
The website also walks visitors — “citizen scientists” — through a tutorial on how to identify animals in the images. To help you decide whether an animal is a honey badger or a hyena, Snapshot Serengeti asks you to classify specific features — stuff like curly horns, stripes and bushy tails.
Almost 28,000 citizen scientists volunteered to classify 1.2 million images — a small subset from the ever growing collection. After each image was classified multiple times by citizen scientists, researchers developed a computer algorithm to store their guesses. In other words, an animal would be considered a cheetah if a majority of the viewers classified the animal as a cheetah. Five field experts were also asked to view and classify several thousand images. Researchers found that 96.6 percent of the answers from the citizen scientists agreed with evaluations from the field experts.
This isn’t the first citizen science success project. Zooniverse hosts a variety of scientific projects asking for public participation. Anyone with an internet connection can help explore the surface of the moon or identify sea creatures. These crowdsourced science projects are great for tackling huge data sets and raising awareness about ongoing research. Margaret Kosmala, a researcher at Harvard University and team member of Snapshot Serengeti, explains how the project was able to “take on a life of it’s own in a very cool way. Citizen scientists put together an in depth bird guide for other volunteers.”
We have much more to learn from the Serengeti snapshots. Data from the images also include time and location. Now that the researchers have accurately identified the animals, they can ask how these animals interact over time and in space. The same dataset is also being used to help develop software to detect and classify species in images. The catalog comprises photos captured from 2010 to 2013 and is featured today in the journal Scientific Data.