OXI: An online tool for visualization and annotation of satellite time series data

Satellite telemetry data is a special case of multivariate time series characterized by large volumes (in terms both the number and samples), varying sampling rates (including gaps), redundant sensors, many interconnections between series. Special tools are needed to handle visualization, analysis, and, most importantly, annotation such data. Manually prepared annotations crucial when designing evaluating algorithms for satellite including anomaly detection in Although there applications no that would smoothly typical at same provide free user-friendly interface accessible from any computer without need installation, registration, sharing, or training. As solution this technology gap, we propose web-based OXI tool written JavaScript publicly available https://oxi.kplabs.pl/. Based on TRAINSET application (https://trainset.geocene.com/), optimized each part real European Space Agency (ESA) missions. The important improvements include redesigned loading visualization effectively datasets (up 2,000,000 different modes (1D/2D single/multi-series), autosaving status browser storage, dozens user experience enhancements (autoscaling axes, color-coded selectors, additional hotkeys, others). was designed evaluated cooperation with domain experts Airbus ESA. Finally, an extensive description software, design choices, their impact creating machine learning models telemetry.