Tipping and Cueing of Live Full Motion Video

We propose applying Matroid’s solution to provide nonhuman tipping and cueing for full-motion video to enhance the Air Force’s current processing, exploitation, and dissemination (PED) process. We’ve developed an intuitive UI and automated infrastructure to empower non-programmers to create high accuracy detectors and easily deploy them across live HD video streams and/or historical video. This means that Imagery Analysts, with no background in programming, can gain assistance from state-of-the-art computer vision models and massively scalable, performant machine learning pipelines to detect anything they want in video. No other company has done this to date. This power enables imagery analysts and other PED team members to benefit from the cutting-edge progress that Computer Vision and AI has recently achieved. Solving this manpower bottleneck for Analysts may be the single most important factor in supporting the huge growth in demand for unmanned ISR feeds.