June 11--Images from "camera traps" set up in Serengeti National Park in Tanzania offer a candid -- and sometimes amusing -- glimpse into life on the African plains.
As a doctoral student, ecologist Alexandra Swanson wanted to observe how the Serengeti's predators interacted with other species. So in 2010, she set up 225 camera traps, or automatically triggered cameras, around the park. Then she spent years driving from camera to camera, changing their memory cards and batteries every two months.
In the process, she captured photographs of some of the Serengeti's most famous fauna, including lions, cheetahs and zebra, as well as the lesser-seen honey badger, zorila and aardwolf.
"When we first started getting these photos they were just breathtaking," Swanson said. "You're seeing these animals at their most authentic.
"You just see things you'd never otherwise see, these animals making ridiculous faces, peering into the camera, running toward the camera."
Camera traps allow researchers to observe wildlife in remote locations or to monitor species as they move across vast areas.
Even in Los Angeles, scientists are using camera traps to track a cougar, and perhaps others, in Griffith Park.
The trade-offs are clear, Swanson said. The automatically triggered cameras are cheap and not very invasive to wildlife (though every once in a while, an elephant might smash one or a hyena might eat one). However, they produce an impossible number of images and there's no easy way to process them.
"A lot of people who use camera traps are often overwhelmed by the number of photos they have to go through," Swanson said.
In three years, Swanson's study amassed 1.2 million sets of images, which were published and described in the journal Scientific Data this week.
Combing through and analyzing such a large number of images was too much for one person. At first, she tried to recruit a dozen undergraduates, but she still couldn't keep up.
Swanson asked fellow ecologist Margaret Kosmala, whose background is in computer science, if there was a way for a computer to do it.
"I said no," Kosmala said. "Computer vision research isn't actually there yet in terms of identifying animals in pictures."
But then Kosmala saw the images and remembers being blown away. She thought maybe they could recruit volunteers, hundreds of them, to help.
Swanson and Kosmala teamed up with Zooniverse, an online citizen-science platform, to create Snapshot Serengeti.
Since its launch in 2012, 30,000 people have logged on to the project's website to help make 10.8 million classifications. The volunteers found animals in more than 300,000 images, and identified 40 different species.
"We literally couldn't have gone through all of the pictures without the volunteers," Kosmala said. "This is science that couldn't have happened without them."
With no active recruiting, Snapshot Serengeti attracted volunteers from Zooniverse's existing following and early press coverage. Zooniverse began as an astronomy-based platform, asking volunteers to help identify the shapes of galaxies on a project called Galaxy Zoo. The site has since expanded to 42 projects across the disciplines, including Penguin Watch and Plankton Portal.
Anyone, even nonexperts, can participate.
"The platform is designed in a way that whether or not they're an expert, anyone can make a contribution," Swanson said. "It doesn't matter if you have no idea what you're looking at."
Snapshot Serengeti presents users with a camera trap photo and asks users to choose from 54 types of animals, including birds, reptiles, insects and even humans. Sometimes there's nothing there.
If you can't tell, you can narrow the search by characteristic -- what it looks like, color, pattern, horn shape. You're also asked to report other details, such as how many animals there are and what they're doing.
How do the scientists know if their volunteers are correct? The site employs a voting system where multiple people are shown the same image.
"If 10 people say it's a zebra, then we know it's a zebra," Kosmala said. "But if many people say it's a zebra and one says it's a giraffe, then we still know it's a zebra."
When Swanson and Kosmala compared the citizen scientists' results to an expert's identifications, they found their volunteers were correct 97% of the time. That's impressive, Swanson said, especially once you consider that even experts make mistakes.
Swanson said she hopes the images can be used for further ecological research and education. Scientists can view the images for their own studies on the Serengeti or to compare that region's wildlife with other areas.
But a greater promise comes from helping to improve computer vision research -- training computers to recognize animals. To teach a computer to recognize a cheetah, scientists need large sets of images with the subjects already identified.
Some image sets like this already exist, but most are well-composed and well-lit -- two things hard to re-create with remote cameras. Camera traps often capture images with the critter out of focus, only partly in the frame or with other individuals -- and computer programs must be trained to account for all these factors, researchers said.
To search the whole image database, go to talk.snapshotserengeti.org/#/search.
"Like" Los Angeles Times Science Health on Facebook.
