DeepDigitalForest
Total project duration:
3 Years
The project
The aim of the project was to develop a cost-efficient workflow for the digitisation of the forest, ranging from the acquisition of airborne remote sensing data of medium to very high point density and its processing, to optimised field surveys using terrestrial laser scanning, to a tablet application to support forest management.
Our activities in the project
JR's task in the project was to develop deep learning (DL) methods for recognising individual trees from LiDAR data with a high point density.
Research groups
Downloads
Auftraggeber
Österreichische Forschungsförderungsgesellschaft FFG
Project partners
UMWELTDATA Gesellschaft m.b.H. (Koordination)
AeroMap GmbH
E.C.O. Institut für Ökologie Jungmeier GmbH
Souveräner Malteser-Ritter-Orden Waldbetrieb Ligist
Project details
In the DeepDigitalForest project, the concept of a consistent, cross-company and cost-efficient workflow for digitising the forest is being expanded by the acquisition and evaluation of flight strips with very high-resolution data (VHR-ALS, low-altitude flight strips). These enable a more precise description of individual trees and must be integrated into the existing system of aeroplane ALS and terrestrial surveys. In addition, a method patented by Umweltdata GmbH for terrestrial laser scanning in the forest is to be integrated into the workflow.
Another focus is on advanced AI-based extraction of tree properties, on expanding the tablet app for complex issues and very large planning areas and on obtaining indicators for monitoring structural richness and biodiversity in forests. Overall, the cost-efficient application of the methods in very large regions or entire nations is to be promoted and established as an international standard.
