Multimedia Sensor Networks (MSNs) have enhanced the ability to analyze the environment and provide responses based on its current status. Generally, MSNs are composed of scalar and multimedia sensors that have fixed locations. However, given the advancement of smart mobile device technologies, it is currently possible to dynamically integrate mobile sensors into MSNs. In this paper, we propose a formal platform to manage MSNs and the data gathered from them to detect complex events. Our main contributions include: M2SSN ​− ​Onto, a Mobile and Multimedia Semantic Sensor Networks Ontology; Py-CEMiD, an engine for detecting complex events and generate reactions to them; a mobile device location engine to locate mobile sensors; and a proof-of-concept in the context of detecting emergency situations in smart buildings. Several scenarios are validated for emergency events, combining simulated sensor measurements with real measurements of mobile devices. Results show complex events can be detected in near real time (less than 1 ​s).