Mapping Marine Macroalgae along the Norwegian Coast Using Hyperspectral UAV Imaging and Convolutional Nets for Semantic Segmentation
- Resource Type
- Conference
- Authors
- Skjelvareid, Martin H.; Rinde, Eli; Hancke, Kasper; Blix, Katalin; Hoarau, Galice G.
- Source
- IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium Geoscience and Remote Sensing Symposium, IGARSS 2023 - 2023 IEEE International. :4002-4005 Jul, 2023
- Subject
- Aerospace
Components, Circuits, Devices and Systems
Fields, Waves and Electromagnetics
Geoscience
Signal Processing and Analysis
Semantic segmentation
Imaging
Geoscience and remote sensing
Forestry
Autonomous aerial vehicles
Spatial databases
Convolutional neural networks
- Language
- ISSN
- 2153-7003
Marine macroalgae form underwater "blue forests" with several important functions. Hyperspectral imaging from unmanned aerial vehicles provides a rich set of spectral and spatial data that can be used to map the distribution of such macroalgae. Results from a study using 81 annotated hyper-spectral images from the Norwegian coast are presented. A U-net convolutional network was used for classification, and accuracies for all macroalgae classes were above 90%, indicating the potential of the method as an accurate tool for blue forest monitoring.