UAV-Based Automated Labeling of Training Data for Online Water and Land Differentiation
- 632 Downloads
Wetlands scientists require detailed maps of the land and water contours to correctly understand wetlands water systems. Current methods rely on expensive, manually collected data that are rarely updated, providing obsolete data for modeling. Automating the process would improve data collection and enable scientists to improve wetlands models. The automated solution needs to minimally interact with the fragile environment, to function autonomously to avoid long trips to remote wetland locations, and to provide modularity to support the different measurement techniques for topographic and bathymetric mapping. Unmanned aerial vehicles with varying payloads and autonomous behavior provide a reasonable and effective solution.
We are grateful to USDA-NIFA 2017-67021-25924, and NSF CNS CSR-1217428, which partially supported this work. We also would like to thank Najeeb Najeeb, Zachary Shields, Nicholas Vaughn, and Tristan Watts-Willis for their assistance.
Supplementary material 1 (mp4 34307 KB)
- 2.Ramachandra, T.V., Kumar, U.: Wetlands of greater Bangalore, India: automatic delineation through pattern classifiers. Electron. Green J. 1 (2008) Google Scholar
- 3.Martyn, R.D., Nobel, R.L., Bettoli, P.W., Maggio, R.C.: Mapping aquatic weeds with aerial color infrared photography and evaluating their control by grass carp. J. Aquat. Plant Manage. 24, 46–56 (1986)Google Scholar
- 8.Xu, A., Dudek, G.: A vision-based boundary following framework for aerial vehicles. In: Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (2010)Google Scholar
- 9.Basha, E., Watts-Willis, T., Detweiler, C.: Autonomous meta-classifier for surface hardness classification from UAV landings. In: Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Vancouver, Canada (2017)Google Scholar