Computer-implemented method, computer program, storage medium and system for video surveillance

A method comprising: querying, based on a first video stream from a first camera, a knowledge graph representing a plurality of cameras as ontology units in order to identify a second camera, and identifying a second program into which the video stream from the second camera may be input to generate second data. The program may comprise object or event recognition. The second program may be launched based on a determination that the object has or is about to disappear from the first video stream. The second camera may be identified based on a determination of which camera FOVs the object can reach within a predetermined time. The method may be enacted by a video management system. The system may provide an alarm or make a log entry in response to the object moving between camera FOVs.