We at HeiGIT together with the group GEORISKS at the German Aerospace Center (DLR-DFD-GEO) are currently offering a Master thesis in the context of Machine Learning for characterizing building inventories related to natural hazard risk models. HeiGIT is a non-profit organization with the objective to improve knowledge and technology transfer from fundamental research in geoinformatics to practical applications. The GEORISKS Group at DLR develops thematically relevant information products based on earth observation data with the goal to support the entire disaster management cycle.
Natural hazard risk models internalize information about the hazard, the exposed elements, and the corresponding vulnerability, respectively. Thereby, building inventories are an essential component of risk models and describe elements that are endangered by a hazard (such as an earthquake) and susceptible to damage. The associated vulnerability characterizes the likelihood of experiencing damage (which can translate into losses) at a certain level of hazard intensity. Frequently, the compilation of building inventories with both a very high thematic and spatial detail is the costliest and at the same time very much needed component (in terms of time and labor) of risk assessment procedures.
Consequently, the aim of the thesis is to describe building inventories based on the combination of ubiquitously multispectral satellite imagery and vector data (OSM building footprints). From a methodological perspective, we want to implement a set of models in a comparative and innovative manner (i.e., dedicated feature calculation procedures, post-classification methodologies, Graph convolutional network, etc.) for the city of Cologne.
Expertise (may be beneficial but is not a requirement):
- Programming skills (Python, R)
- Learning state-of-the-art machine learning models
- Multimodal perspectives describing the built environment
- Flexible in terms of time
- Insights into working with GIS applications
- University-related work for a charitable purpose
We are happy to answer any further questions you may have. Please contact Anne Schauß (firstname.lastname@example.org). We are looking forward to your application!