Not so long ago, if a biologist wanted to know how an animal behaved, that biologist had to commit hours of their time to watching that animal or collecting evidence and making educated guesses. Imagine how our collective knowledge might have improved with teams of people watching these animals every second of every day.
Asking total strangers to help with research might seem strange, but that is the theory behind crowd sourcing through the Internet and the subject of a new citizen science project, Wildlife@Home.
The project, "Integrating Crowd Sourcing, Volunteer Computing and Expert Observation to Robustly Classify Massive Quantities of Avian Nesting Video," was developed by Susan Felege, an assistant professor in the Department of Biology, and Travis Desell, an assistant professor in the Department of Computer Science.
New camera technology is allowing avian ecologists to perform detailed studies of avian behavior, nesting strategies and predation in areas where it was previously impossible to gather data. Unfortunately, studies have shown mechanical triggers and a variety of sensors to be inadequate in capturing footage of small predators (e.g., snakes, rodents) or events in dense vegetation. Because of this, continuous camera recording is currently the most robust solution for avian monitoring, especially in ground nesting species.
However, continuous video footage results in a data deluge as monitoring enough nests to make biologically significant inferences results in massive amounts of data, which are unclassifiable by humans alone.
According to Susan Felege, the task would take years relying on student viewers alone. "We collect between 30,000 and 50,000 hours of video footage a year. Even with a large team of undergrads assisting in the first 30,000 hours of grouse video collected in 2012, we have only reviewed about 1/3 of it over the past year with additional data on two more species collected in 2013."
Currently the project is looking at video of sharp-tailed grouse performing their mating dances and then examining their nesting habits and ecology. The nest cameras have been set up near western North Dakota's oil fields and also within protected state lands.
"The videos we are collecting will shed light on impacts of gas and oil on grouse nesting success, predator interactions with grouse at the nest, and nesting behaviors such as foraging," says Felege. "In addition, data collected on piping plovers and least terns will assist us in accurately determining nest fates used in conservation efforts of these federally listed species, as well as impacts of researchers on nesting areas, nesting ecology, and nest predation."
All of this information should help avian ecologists better understand nesting ecology and ultimately make better management recommendations for the species. Through the citizen science project, researchers have the opportunity to educate the public and provide opportunities to get the public excited about avian conservation efforts that otherwise would be difficult for them to experience.
Handling all this video data is an immense challenge for researchers, so the team looked to volunteer computing, a form of distributed computing where people share their personal computer's data processing and storage space. In this case, the video is stored on UND servers, but volunteers view videos on their personal computers and volunteer their interpretive skills to assess what is happening in a given video.
Volunteer computing is a part of crowd sourcing and gives people the ability to participate in real research in a way unimaginable just a few years ago. According to Desell, "Volunteer computing gives people a real democratic voice in terms of the science they support. They vote with their computers by using them on the projects they think are most important. Not only that, but it really gets the public involved in citizen science, where they can actively participate and engage in research on different scientific projects. I think it's a great way to get future generations involved in the STEM (science, technology, engineering, and mathematics) fields."
Volunteers can go to the site, register, and begin watching videos quickly, making a meaningful scientific contribution in less than ten minutes. To make sure researchers get as much quality data as possible, each three minute video is watched by more than one volunteer, and a comparison is made. If one viewer says he saw an act of predation, but another viewer does not confirm this, then the video is flagged and a student watches the video to determine what happened.
And just to make the process more enjoyable, Desell and Felege added a credit tracking system, where users are awarded points for watching video, or running computer vision algorithms on video. Leaderboards track which users have processed the most video and watched the most video. Users can also collaborate in teams, and there is another set of leaderboards to see which teams have cumulatively done the most.
By all accounts, Felege and Desell seem to have succeeded at creating a model for incorporating the work of average citizens in meaningful research on endangered avian populations. To date, volunteers have reviewed thousands of hours, but there is still work to be done. And that work can be done by anyone with a home computer. A&S
Arts & Sciences
Two UND professors have come up with an innovative way to do research that include thousands of hours of video.