Marine Debris Program Partners to Develop Prototype Application to Locate, Classify, and Count Large Marine Debris
JAN. 10, 2022 — As part of a Fiscal Year 2020 award from the NOAA Office of Marine and Aviation Operations Office of Uncrewed Aircraft Systems Program, researchers from the NOAA National Centers for Coastal Ocean Science (NCCOS), Oregon State University, Genwest, and ORBTL.AI, together with the NOAA Marine Debris Program, have developed a publicly available prototype application for classifying marine debris from aerial images that users upload.
The application is hosted by NCCOS on GitHub and was developed by training a deep learning-based object detection model using 5,733 hand-labeled marine debris objects spread across 1,224 aerial images collected in Hawaii in 2015. If applied correctly, machine learning (and the subfield of deep learning) are a powerful technology for automating tedious tasks previously performed by hand including the identification of shoreline marine debris.
For more information, please contact Amy.Uhrin@noaa.gov.