Vidrovr Inc. SBIR Phase I Award, December 2019

Vidrovr believes current computer vision solutions are incomplete and do not provide a tractable solution for the Air Force in use cases pertaining to target reacquisition or re identification for FMV. This problem can be reformulated as a visual search problem. No off the shelf solutions provide visual search on video. There are many issues around this problem domain - lack of annotated training data as well as fast model adaptability of for out of scope classes comes up often. These are some of the machine learning problem that Vidrovr's dual use systems aim to solve in order to deliver visual search for FMV.