Determine plant coverage with image segmentation – reindeer lichens
Measuring coverage of different plants or plant groups is an important part of large national inventory programs such as the national forest inventory (NFI) and the National Inventory of Landscape in Sweden (NILS). Data are collected continuously to describe the state of nature and its changes. A fundamental problem with current coverage estimation is that there is a variation up to 20% between different field workers. This affects the ability to detect real changes. We believe that developing of image segmentation to assess coverage of different vegetation types in photos taken during field inventory using machine learning can reduce person variation. In the project, the students would work on developing existing image analysis programs and improve them to fit the identification of some different plant and/or vegetation types that are found in the forest and which are important reindeer plants in the winter.