For more than a decade tracking and tracing physical objects has been target of information systems within the realm of research on the Internet of Things. But application to human populations requires reconsideration of re-identification and data locality requirements due to ethical and legal constraints. For this domain, we propose a generic event recognition architecture (GERA) and evaluate its applicability for developing a sensor-based information system for recognizing moving population densities by obeying non-re-identification and data decentrality requirements. Empirical evaluations show that this information system provides mean structures for measuring event data and deriving predictions that are statistically equal to manually measured actual data. Finally, a general discussion on the integration of event recognition systems into busi-ness process environments is given.