This chapter presents an interactive video-surveillance solution for assisting human operators in the control of movements across a multi-region scenario (perimeter control). It has been conceived as a multi-camera system to detect anomalous trajectory events, such as entering or leaving a region or changing the walking speed, by means of a dynamic collection of decision rules. They relate spatio-temporal patterns and event categories (anomalous, unknown, normal), and are used to assess and classify trajectory events. The interactive paradigm has been adopted as a natural framework to progressively learn and update rules, particularly at early stages of the system operation. The approach of continuously improving system knowledge from user feedback conducts to adaptive, reliable and increasingly automatic systems in a relatively short period of time.