In many developing countries, coastal areas show high dynamics of settlement structures, which are hardly regulated by regional planning and therefore give rise to a series of risks. Most of all, increasing settlement density and spread in areas close to the shoreline and into wetland areas appear worrying against the background of climate change and sea level rise. In our study area, the coastal zone of Benin, settlements are spreading into agricultural areas as well as near-natural zones, towards the lagoon and are threatened by coastal erosion. To enable regional planners to take these threats into account, process monitoring respectively modelling based on remote sensing data is needed. Due to very small land use structures and the necessity of detecting individual buildings, very high spatial resolution (VHSR) data has to be used. However, like in other developing countries, VHSR data availability is poor. Furthermore, process analysis and modelling based on approaches for industrialized countries are not feasible due to strong differences in the appearance of villages respectively suburban areas in Benin. Individual buildings are sometimes even difficult to detect by eye, nonetheless, to achieve large-scale information, automation is indispensable. We exemplify these issues for the coastal area of Benin by an approach based on both manual and segment-based (semi-)automatized building detection. We use the results to analyze the settlement process and model its further evolution by data driven modelling.