Camera traps, which are motion- or heat-activated automatic cameras, are revolutionizing how researchers can study ecosystems, since they are noninvasive, relatively inexpensive, and are capable of monitoring large areas and diverse species. A few years ago, members of the research group of MSI Principal Investigator Craig Packer, a Distinguished McKnight University Professor in the Department of Ecology, Evolution, and Behavior (College of Biological Sciences), set up a camera-trap survey in Africa to study the way predators and their prey co-existed. The study area included 225 camera traps in a 1,100 km2 region in Serengeti National Park in Tanzania.
The world’s largest camera-trap survey, this project generated far more images than the research team was able to identify on their own. Then-grad students Alexandra Swanson (now a post-doc at the University of Oxford) and Margaret Kosmala (now a post-doc at Harvard University) decided to partner with the citizen-science web platform Zooniverse.org to create Snapshot Serengeti, a project that allowed members of the general public to look at the images and classify them. Over 28,000 volunteers participated in the project. Participants were able to identify what species they thought was in the photo, what the animal was doing, whether there was a young animal present, and several other data points. Each image was classified by multiple participants, and an algorithm created by the researchers collected the responses and identified a consensus description for each image. The group has published their research in Scientific Data, which is an online publication of the prestigious journal Nature (Alexandra Swanson, Margaret Kosmala, Chris Lintott, Robert Simpson, Arfon Smith, and Craig Packer. 2015. Snapshot Serengeti, high-frequency annotated camera trap images of 40 mammalian species in an African savanna. Scientific Data 2:150026. 10.1038/sdata.2015.26). The research group hopes that the dataset they’ve collected will be useful for future research and education.
This paper received a huge amount of media attention when it was published in June. Media outlets that have published the stories include the New York Times, the Washington Post, the Los Angeles Times, PBS Newshour, BBC News, the Daily Mail, Scientific American, Discover Magazine, Wired, the New Yorker, major newspapers in Brazil and India, and many others. A story also appeared in the University of Minnesota Discover blog.
From 2010 through 2013, the project generated 1.2 million image sets (each set consisted of 1-3 photos), which comprised 4.5 TB of data. This huge amount of data necessitated a large storage capability that MSI has been able to provide. In February 2015, at the suggestion of the UM Informatics Institute, the research team decided to transfer the images to MSI’s new Ceph object storage system. This system has proven to be an excellent resource for the researchers, particularly since the images can be shared with colleagues outside the University of Minnesota and it has a built-in web serving capability that makes creating sites quick and easy.
MSI has also assisted with another Zooniverse project, Ancient Lives. A Research Spotlight about this project appeared on the MSI website in December 2014.
Image description: The Snapshot Serengeti website interface. This is the primary interface with all available species options (left) and filter that help narrow users’ choices when classifying species (right). Image and description from Swanson et al., Sci. Data 2:150026 DOI:10.1038/sdata.2015.26 (2105).
posted on July 8, 2015