Using Artificial Intelligence for Cost-Effective Environmental Monitoring and Site Characterization
Scientists at Integral Consulting are “training” computers to recognize and measure physical and biological features in images of the seafloor to help map benthic habitats found beneath the sea. Over the last two years, Integral’s research team has compiled curated image sets containing hundreds to thousands of labeled examples of organisms, including worms, shrimp, and urchins, a full range of bottom textures, and biogenic structures such as feeding pockets and burrows.
Supplied with these training sets, Integral is using established computer vision and pattern recognition methods to identify these features and automatically extract corresponding information from newly collected images. This automated image processing allows Integral scientists to generate large, accurate data sets from seafloor image surveys in a fraction of the time it would take using manual image analysis methods.
The research is part of work supported by the U.S. Department of Energy’s Office of Renewable Energy and Energy Efficiency under the Water Power Program Office Award Number DE-EE0007826. Senior sediment scientist, Gene Revelas, and oceanographer and remote sensing expert, Brandon Sackmann, Ph.D., are heading up this project to develop a streamlined and standardized seafloor habitat mapping approach using well-established benthic mapping survey techniques such as multibeam echosounder acoustic and sediment profile imaging (SPI) surveys combined with state-of-the-art computer vision approaches for rapid data and seafloor habitat map generation.
“The recent advances in artificial intelligence provide significant opportunities to automate and standardize the generation of environmental data from underwater, or for that matter, any imagery,” explains Dr. Sackmann. “Our goal is to bring the analysis and data extraction process associated with tried and true high-resolution underwater camera systems, such as SPI, into the 21st century for more cost-effective seafloor characterization and monitoring.”