Real-time Evaluation, Compression, and Organization of Recorded Data using Extensible Rules (RECORDER)

Navy ships have many cameras for situational awareness, but they produce more video than humans can effectively review, and the gap is widening. Currently, all video is recorded until the disk is full and the rest is lost, so the sparse relevant data (e.g., interesting targets or conditions) is buried in the noise or not recorded at all. This lost data compromises decision-making and delays machine learning and autonomy efforts. We developed Real-time Evaluation, Compression, and Organization of Recorded Data using Extensible Rules (RECORDER) to maximize the value of recorded data, adjust the disk contents when it approaches its storage limit, perform selective compression on different pieces of data, and take the dataÆs usage into account when selecting a compression approach. In Phase I, RECORDER showed an ability to compress data between 90% and 97% while preserving orders of magnitude more targets of interest than na’ve approaches and maximizing disk space usage better than ōpositive criteriaö recording. In Phase II, we will develop a full-scope prototype and evaluate it in progressively more realistic demonstrations.