You are here:
Deep Lake Explorer: Using Citizen Science to analyze underwater video from the Great Lakes. (presentation)
Citation:
Wick, M., T. Angradi, M. Nord, M. Pawlowski, R. Debbout, J. Launspach, AND D. Bolgrien. Deep Lake Explorer: Using Citizen Science to analyze underwater video from the Great Lakes. (presentation). North American Lake Management Society, Burlington, VT, November 11 - 15, 2019.
Impact/Purpose:
We will present preliminary results of the beta test and public launch of Deep Lake Explorer, a citizen science application to facilitate analysis of underwater video collected as part of the National Coastal Condition Assessment. Results of video collected in the Niagara River, Lake Huron and Lake Ontario will be presented. This project is one of the only projects to incorporate video footage into this type of analysis, and the first time EPA has used an online citizen science project to help interpret imagery or video. Our presentation will demonstrate the process and benefits of using citizen science to interpret underwater video, and our results will set up the EPA to effectively analyze underwater video collected in future iterations of National Coastal Condition Assessment.
Description:
Deep Lake Explorer is a web application hosted on the Zooniverse platform that allows citizens to participate in limnological research by helping to analyze underwater video collected from the Great Lakes (https://www.zooniverse.org/projects/usepa/deep-lake-explorer). The project is part of a US EPA Office of Research and Development pilot study to develop underwater video as a tool for assessing ecosystem condition as part of the National Coastal Condition Assessment (NCCA). It is one of the first Zooniverse projects to analyze underwater video clips rather than still images. Video clips are analyzed for the presence of invasive species like round gobies and zebra and quagga mussels. The collection of large underwater video datasets presents unique challenges to data analysis, as they are time-consuming to analyze. Interpretation of underwater video is often difficult due to variability in water clarity, bottom complexity, light, and camera movement. Movement of organisms can make them easier to notice, but difficult to identify. Deep Lake Explorer harnesses the power of crowdsourcing to obtain multiple interpretations of each clip, which can then be aggregated for each video to determine the presence of invasive species across the Great Lakes in a quick and cost-efficient way. We have beta-tested the application with video from Lake Ontario, Lake Huron, and the Niagara River, and compared results to expert analysis. Consensus-based results of Zooniverse participants agreed with experts’ results for fish presence for 84% of clips, and for 88% of clips for mussel presence. Additional methods of cleaning and aggregating user data for comparison with expert analysis is being explored to increase the reliability of results. The full project will launch in summer 2019.