With efforts like [email protected] and [email protected] in the spotlight, citizen science is having a moment. Many of these efforts are particularly emphasizing gamers, who tend to have powerful desktop computers and the geeky willingness to dedicate them to scientific causes. Now, NASA is highlighting a new effort that loops iOS and Mac gamers into the mix. The mission: saving the world’s coral reefs.
Coral reefs, which serve as vital habitats for innumerable species, are rapidly deteriorating and dying, leaving scientists scrambling to combat the decline. To understand the problem better, NASA has been deploying a pair of technologies to map the oceans in great detail: FluidCam and MiDAR. FluidCam is a high-performance digital camera that uses software to undo the lensing effect created by water in order to see beneath the surface, while MiDAR collects light bounced back to the instrument – the kind of technology that could allow a satellite to detect a coral reef at a centimeter resolution.
For years, NASA has been using these technologies on drones and aircraft to map the ocean floor around areas like American Samoa, Guam and Puerto Rico, capturing vivid images of and data on coral reefs, seagrass and more. But after years of data collection, one problem remained: data labeling. Without a massive influx of labeling to make the data useful, NASA couldn’t apply high-performance technologies to analyze the data.
To bridge this gap, NASA introduced a new initiative called the Neural Multi-Model Observation and Training Network, or NeMO-Net. But – at least on the surface – NeMO-Net isn’t a piece of academic software: it’s a video game. When playing NeMO-Net, normal people are able to travel the ocean floor in a virtual submarine, helping to classify corals. NeMO-Net even has built-in features like achievements and progress tracking.
The data from NeMO-Net’s playerbase is fed to NASA’s Pleiades supercomputer. Pleiades is an SGI/HPE system, currently ranked 32nd on the Top500 list of world’s fastest computers, with nearly 6 Linpack petaflops of performance provided by 11,207 Intel-based nodes. Using the player-generated data, Pleiades is learning to recognize coral from any ocean floor image, even when FluidCam and MiDAR were not used to take them.
“NeMO-Net leverages the most powerful force on this planet: not a fancy camera or a supercomputer, but people,” said Ved Chirayath, principal investigator for the project and a member of NASA’s Ames Research Center. “Anyone, even a first grader, can play this game and sort through these data to help us map one of the most beautiful forms of life we know of.”
NeMO-Net is currently available for iOS and Mac devices, with an Android release planned for the future. To download the app, click here.