Dismount Tracking in Urban Scenes

This SBIR Phase II project will develop an advanced dismount activity analysis system that will utilize novel statistical models and non-causal techniques for robust dismount detection and tracking in low spatial resolution and frame-rate aerial videos. The key innovation in this effort is an integrated approach for dismount detection and tracking that explicitly models typical scenarios including low contrast and pixel resolution, clutter, occlusions, missed detections, complex target interactions, and grouping behaviors. In Phase I, we demonstrated the proof of this concept by developing core technologies that included i) Bernoulli variable based statistical target detection by jointly modeling spatio-temporal properties of scene background and sensor geometry, ii) deferred inference data association framework for interacting targets, iii) efficient algorithms for optimizing the data association problem, iv) modeling target interactions and grouping behavior for tracking, and v) a forensic video analysis tool for proof of concept. The Phase II effort will be focused towards refinement, advancement, and integration of these enabling technologies and will include development of new algorithms to handle the issues identified during Phase I, development of an integrated detection and tracking system, transitioning of technology and prototyping, and detailed quantitative and qualitative evaluation of the system and its components.