Clemson University Partners with Perceptual Informatics to Fight Forest Fragmentation
Asheville, NC (October 5, 2019) - Clemson University's Life Sciences Outreach Center (LSOC) has partnered with Perceptual Informatics, LLC to develop a new object-detection database to assist AI-powered systems to recognize forest fragmentation in satellite images. The collaboration will deliver both a searchable image database of global high-resolution satellite imagery and a prototype AI model that detects the early signs of forest fragmentation. The project will bring together Clemson students, faculty, the U.S. Forest Service, lay citizen scientists, and industry professionals, and will be executed upon successful funding through the Leonardo DiCaprio Foundation and Microsoft's AI for Earth Innovation Grant.
Fragmentation is not equivalent to deforestation though the two are related. Fragmentation is a process where "edges" are introduced into previously continuous forest through human development, such as roads and agriculture. Edges provide an easy way for invasive species, disease, and human induced wildfire to gain entry to a forest. It's a little like an injection, except that it's continuous. Fragmentation leads to sharp declines in the biodiversity and abundance of interior forest species, and often inadvertently leads to population booms in edge dwelling species like deer and raccoons.
The Fight Forest Fragmentation project is a pilot study to see if lay citizen scientists can help ecologists and foresters scan and label thousands of satellite images for signs of forest fragmentation. Perceptual Informatics will help the project develop the image labeling tool as well as obtain the Synthetic Aperture Radar data that will be used to visualize forested landscapes. The final database of images will be used by object-detection AI algorithms to identify new locations of forest fragmentation around the world on their own and alert scientists. This new database and system will also complement existing forest fragmentation databases and ongoing global forest monitoring systems because it is the first of its kind to use all-weather, day-and-night radar data from satellites combined with AI.
Dr. Renee Lyons, Clemson University's Director of Science Outreach, will serve as the project's Principal Investigator and will provide opportunities for students and members of the public to contribute to the development of the AI-system. The project will teach students not only how to recognize forest fragmentation but also its effects on the quality of forest habitat, ecosystem services, and what this means for shared cultural values. The project will also provide students the opportunity to learn cutting-edge skills in artificial intelligence and computer science. Perceptual Informatics will lead the technical execution of the project and will ensure that the database and system are developed in a sustainable and scalable manner.