On November 12, 2026, Marlene Bauer will present her Master Thesis on ” Robust and Uncertainty-Aware Informal Settlement Mapping — Cross-City Generalisation Across Diverse Urban Morphologies Using High-Resolution Satellite Imagery in the Global South ” at 13:00 at the EORC Meeting Room (1st floor) in John-Skilton-Str. 4a.
From the abstract:
Can artificial intelligence map informal settlements in a city it has never seen, and how much can we trust the results? This thesis investigates how well deep learning models can identify informal settlements in satellite imagery across 12 cities in Africa, Asia, and Latin America, where urban structures and settlement patterns vary considerably. Each model is tested on a city it has never encountered during training, revealing how well it generalises to new urban contexts. Beyond identifying informal settlements, the models also estimate the likelihood of each prediction and its associated uncertainty.
The findings highlight the potential of AI to map informal settlements across diverse cities, even where local data are scarce. By estimating both the likelihood of informal settlements and the uncertainty of these predictions, the approach provides a more nuanced picture of settlement patterns and the reliability of the resulting maps
1st supervisor: Prof. Dr. Hannes Taubenböck
2nd (external) supervisor: Dr. Thomas Stark, DLR
