As sensor networks grow in size and variety, the copious amounts of data become difficult for a user to comprehend. Furthermore, if sensors are mobile, e.g., video cameras carried on board an unmanned aerial vehicle (UAV), a user must also control the sensor platforms. Such control is difficult for a single user, especially as the number of platforms grows. We present a system that addresses these issues, focusing specifically on UAV video sensors performing surveillance over large, remote areas. A single user controls a large fleet of UAVs while monitoring the sensor data by interactions with a multimedia database. Built on a geographical information system (GIS), the database stores and retrieves sensor and platform data. The user controls the UAVs by specifying predefined surveillance task in the database. A multiagent system (MAS) uses information in the database to plan the actions of the UAVs as they conduct their surveillance tasks under changing conditions. Predefined surveillance tasks allow for minimal communication between agents and reduce the computational power required by each UAV. We demonstrate the system using a small fleet of simulated UAVs collecting images over a surveillance area.