Abby-Net
Over the last couple of years, the number of earth observation data drastically increased as new satellite missions emerged and sensor equipped aerial vehicles became more and more common. Although there exist multiple products which can be derived by physical models from raw sensor data, the use of modern AI technology can open up new applications, improve data quality and significantly reduce the effort to analyse raw sensor data. In this talk, we will discuss the use of remote sensing technology for research in sustainable energy transitions. After highlighting the general properties of Deep Learning, we will focus on typical problems of this machine learning paradigm in remote sensing applications. Finally, we will highlight several novel AI methods that can help to overcome these challenges. In particular, the talk highlights methods in the areas of spatial interpolation, super- resolution networks and label generation in weakly supervised settings.
Source
“Surveilled PlanetAI-Driven Earth Observation”by Matthias Schubert
