The PANTHER (Pattern ANalytics To support High-performance Exploitation and Reasoning) Grand Challenge officially kicked off on October 1st. This represents an enormous laboratory investment in a project that will represent a new level of integrating capabilities in 1400 and 5000 to bring solutions to some of the most important problems facing the current generation of remote sensing platforms. PANTHER will advance the science of remote sensing to exploit critical patterns and relationships in high-volume, pixel-based national decision systems. These systems detect the subtlest of features, including spatially and temporally distributed cues associated with gradually unfolding events. However, by focusing on pixels, current exploitation infrastructure enables only a fraction of sensors’ full potential − in sensitivity, specificity, and magnitude. Furthermore, this paradigm does not scale to the increase in pixel-based data volumes anticipated in coming decades. To maximize remote sensing capability, scientific advances are required across three key technical domains: Sensor Exploitation, where innovative algorithms extract and characterize relevant entities from raw data streams in real time; Discrete Analytics, where semantic graphs and tensors represent, index, and correlate entities, enabling complex spatiotemporal queries; and Human Analytics, where empirical studies of visual cognition inform technologies that support rapid, data-based situational assessment via pattern-based content query, retrieval, hypothesis formation, and testing.
November 1, 2012