FIXATE: FIXational Automatic Target Extraction

During a mission naval UAS collect large volumes of data but only sub-sections of this data contain information relevant to the mission. In order to find the important information analysts must sift through ALL of the data. FIXATE is a biomimmetic object classification system to automatically pre-process the data and cue the operator to sections of data that are likely to contain important information. This means that analysts can focus on the task at hand rather than manually searching hours and hours of video data. Statistical pattern matching techniques and standard computer science based machine learning have achieved great things, however they have had very limited success with object classification with UAV sensor data. FIXATE builds upon an understanding of how the human and primate visual system processes visual information as a series of fixational eye-movements across objects of interest in order to rapidly detect and classify those objects. SSCI has a track record of delivering biologically derived machine learning systems to DoD through the EVASE SBIR program which is currently being deployed at Edwards Air Force Base. FIXATE builds upon how EVASE cues flight test engineers to sections of flight test video containing important information through customizable pre-processing, and updates it for the purposes of Naval ISR data analysis.