This Small Business Innovation Research (SBIR) Phase I research project proposes the unsupervised extraction of contextual information from dedicated video surveillance cameras by providing a mechanism to manually inject pieces of semantic, using semantic knowledge as seed information to learn statistical context models for: functional components, Environmental components, targets, temporal components, and integrating the learned contextual information into video surveillance systems.<br/><br/>Automatic exploitation of context from video will benefit: core computer vision research areas such as segmentation, background-modeling, tracking, and classification, event detection, automatic unusual behavior detection; abnormal target behavior, weather conditions, sequences of events, and the like and event handling. This project will have the most impact in the homeland security and law enforcement areas.