Multi-Camera Real-Time Feature Recognition, Extraction&Tagging Automation (McFRETA)

ABSTRACT: With the proliferation of video sensors in the battle space, Intelligence Analysts are inundated with massive amounts of data, a small proportion of which may contain items of operational significance. EOIR Technologies, Inc. (EOIR) is proposing to develop and deliver a prototype open source, real-time video event detection, metadata tagging and video display system, with search capabilities. This integrated system will leverage an existing video processing and exploitation framework and user interface developed by EOIR, to reduce technical risk. The Phase I research effort will involve: extending the video processing framework for real-time multiple-camera data ingestion; integration of an anomalous event detection algorithm into the framework; adaptation of a metadata standard for event representation; and user interface extension for real-time, multiple-camera data display and rules-based metadata queries. The Phase I product will include the prototype system, as well as the test and evaluation results of both the components and integrated system. BENEFIT: The EOIR solution will provide Intelligence Analysts with a real-time video event processing system based on a tested, open-source video framework, with multiple camera video ingestion, as well as the ability to produce and search standardized metadata tags of anomalous events. Technologies developed under this SBIR are applicable to DoD, Intelligence, and the Department of Homeland Security missions, as well as to commercialization in the private security sector.