Project Overview
Tactical Analysis of Video Imagery (TAVI) is a Scientific Research Group project designed to develop a terrorist threat-detection system based upon surveillance video.
A three-year, $4 million competitively-won award, TAVI will develop and deliver an automated system to provide warfighters with highly-localized, near-real-time intelligence on abnormal terrorist activity, as well as the organizational structure of local terrorist cells.
TAVI is funded by the Office of Naval Research. The WVHTC Foundation is the prime contractor on the project.
Goals and Objectives
When complete, the TAVI system will not only provide expeditionary forces with near-real-time identification and localization of terrorists, but will also provide location-specific intelligence on uprisings and the organization of terrorist cells. TAVI holds great promise in improving intelligence-gathering capabilities.
The project will leverage leading video object-detection and face recognition technologies from ObjectVideo and Identix with proven complex social network analysis techniques developed at Notre Dame University and Carnegie Mellon University, as well as recent developments in network robustness theory at Arizona State University.
In the Global War on Terror, complex social network analysis has been successfully applied to email and telephone traffic, both to identify abnormal levels of activity that might be precursors to a terrorist attack and to deduce the hierarchical structure of terrorist organizations.
While providing valuable intelligence on the global scale, such analyses are often slow and not location-specific, and they are of no use if the terrorists in question do not use these easily-detectable modes of communication.
Recent advances in human figure detection and face recognition from video imagery have for the first time opened up the possibility of applying these same complex social network analysis techniques directly to human activity data extracted from surveillance video.
Such analyses could enable detection of abnormal activity levels at a local level that might provide timely warning of an attack in a specific location, as well as providing the organizational structure of local terrorist cells, complete with images of the cells’ members.
|
