Mask R-CNN in GRASS GIS
The goal of the diploma thesis is a design of a software tool allowing the user to train his own Mask R-CNN model and to apply it to get vector masks from raster data. The library for these tools is heavily based on Python implementation of Mask R-CNN by Waleed Abdulla, Matterport, Inc. (see here).
Bc. Ondřej Pešek
Ing. Martin Landa, Ph.D.
Moritz Lennert, Université Libre de Bruxelles
June 2018
This repository is freezed by the term of the defence. The follow-up development can be found in repository https://github.com/ctu-geoforall-lab/i.ann.maskrcnn.