Making sense of massive amounts of video
Advances in sensor technology from micro-UAVs (unmanned aerial vehicles) to Wide Area Motion Imagery (WAMI) sensors such as ARGUS-IS provide analysts with massive amounts of motion imagery that simply cannot be monitored by human operators alone. The true value of these assets cannot be realized without automated systems that can convert the massive data streams from the sensors into actionable intelligence.
Under several Department of Defense (DoD) funded projects, ObjectVideo is developing technologies and systems focused towards harnessing this data and providing automatic video exploitation, search/retrieval and management capabilities to analysts and warfighters. These technologies include automatic target detection and tracking, event detection, alert generation, automatic target hand-off from one asset to another, georegistration, 3D reconstruction and query and retrieval. This work places new operational tools in the hands of warfighters and analysts for targeting, mission planning, activity analysis of high-valued targets, real-time monitoring and forensic data analysis.
Video Big Data
Continual advancements in technologies are producing sensors that generate exponentially increasing volumes of data covering expanding regions of interest. This, combined with the plethora of commercial and public data sets available, is overwhelming modern warfighters as they try to make sense of this massive amount of information in a timely manner.
ObjectVideo is addressing this challenge with research into advanced analytics algorithms and visualization techniques that present the data in an intuitive and meaningful way. Solutions have been developed to enhance situational context of geo-registered video by overlaying customizable reference metadata highlighting objects and regions of interest in real-time. Simultaneous Localization and Mapping (SLAM) techniques are employed to recover local 3D map and sensor poses from distributed and independent sensors to fuse into a centralized team view of the region. Results are overlayed onto each individual team member's video feed, enabling immersive visualization with user-controllable interface that enhances situational awareness, collaborative engagement, and target-to-shooter handoff.
ObjectVideo's research incorporates Web 2.0 technologies combined with industry media standards to provide light-weight client solutions requiring minimal deployment and maintenance costs and are compatible across a variety of computing platforms.