With the proliferation of inexpensive cameras, video surveillance applications are becoming ubiquitous in many domains such as public safety and security, manufacturing, intelligent transportation systems, and healthcare. IP-based video surveillance technologies, in particular, are able to bring traditional video surveillance centers to virtually any computer at any location with an Internet connection. Today's IP-based video surveillance systems, however, are designed for specific classes of applications.  For instance, one cannot use a system designed for incident detection on highways to monitor patients in a healthcare facility.  To support rapid development of video surveillance applications, we designed and implemented a new class of general purpose database management system, the live video database management system (LVDBMS).  We view networked IP cameras as a special class of storage devices, and allow the user to formulate ad hoc queries expressed over live video feeds.  These continuous queries are processed in real time using novel distributed computing techniques.  With this environment, the users are able to develop various specific web-based video surveillance systems for a variety of applications.  These systems can coexist in a unified LVDBMS framework to share the expensive deployment and operating costs of the camera networks.  Our contribution is the introduction of a live database approach to video surveillance software development.