The University of British Columbia

Published purpose

Sustainable forest management requires high-quality information on tree species, biomass and growth rates. Currently, the ability to accurately map these attributes over forested landscapes accurately and in a cost-effective way is limited not by data, but by processing approaches to extract these attributes from very large, complex datasets acquired from a range of remote sensing platforms. This Project will apply deep learning algorithms to a wide range of remote sensing and other geographic data to predict these critical attributes. It will exploit the key benefits of each of the particular technologies, using algorithms currently not well investigated in a forestry context. The Project will develop a deep learning toolkit, available to forest companies, and academics, demonstrating the appropriate use and application of these new tools to extract these critical forest attributes.