The integration of information acquired with different modalities, spatial resolution and spectral bands has shown to improve predictive accuracies. Data fusion is therefore one of the key challenges in remote sensing. Most prior work focusing on multi-modal fusion, assumes that modalities are always available during inference. This assumption limits the applications of multi-modal models since in practice the data collection process is likely to generate data with missing, incomplete or corrupted modalities. In this paper, we show that Generative Adversarial Networks can be effectively used to overcome the problems that arise when modalities are missing or incomplete. Focusing on semantic segmentation of building footprints with missing modalities, our approach achieves an improvement of about 2% on the Intersection over Union (IoU) against the same network that relies only on the available modality.

Software-intensive systems often consist of cooperating reactive components. In mobile and reconfigurable systems, their topology changes at run-time, which influences how the components must cooperate. The Scenario Modeling Language (SML) offers a formal approach for specifying the reactive behavior such systems that aligns with how humans conceive and communicate behavioral requirements. Simulation and formal checks can find specification flaws early. We present a framework for the Scenario-based Programming (SBP) that reflects the concepts of SML in Java and makes the scenario modeling approach available for programming. SBP code can also be generated from SML and extended with platform-specific code, thus streamlining the transition from design to implementation. As an example serves a car-to-x communication system. Demo video and artifact: http://scenariotools.org/esecfse-2017-tool-demo/