Kiev. Ukraine. Ukraine Gate – January 30, 2021 – Technology
Researchers at the CDISE Center for Scientific and Engineering Computing Technologies for Large Datasets and the Skoltech Space Center have applied a neural network approach to automate the identification of dominant rocks from high- and medium-resolution images.
The use of a hierarchical classification model, as well as additional materials (such as the height of the vegetation cover), made it possible to significantly improve the quality of predictions and ensure greater stability of the algorithm for its practical application.
“Services based on the developed technology can be in demand by commercial companies that perform forest taxation works, the end-users of which are loggers and wood processors, as well as departmental organizations of the forestry industry for quantitative and qualitative assessment of wood resources in leased areas. Our approach can also be used for a rapid assessment of the investment attractiveness of forest areas in underdeveloped forest areas, ”explains the first author of the work, Skoltech graduate student Svetlana Illarionova.
The developed algorithms are planned to be integrated into the Geoalert platform for automating the production of forest inventory materials, implemented using the specialized software Parma-GIS.
Read also: A New Medical Hub in Zaporizhia!