Solution Description:

GameDay enables airmen to subscribe to detected object classes, patterns and behaviors from multi-modal data sources where content is delivered in many specified formats and protocols. GameDay performs real-time PED processing on FMV, WAMI, MTI, aerial and LEO SAR, and EO/IR Imagery. It applies deep learning algorithms to data in parallel to detect targets from live data streams. Detections are disseminated to users and external systems via subscriptions that specify the delivery methods. GameDay algorithms examine many factors based on image or video type ingested, including varying resolutions and look angles. GameDay uses advanced object detection and tracking algorithms to achieve state-of-the-art performance in detecting vehicles, structures, and dismounts with confidence scoring. GameDay uses neural networks to perform image classification, object detection of various types of vehicles, structures, and dismounts, and cell-level instance segmentation of objects. GameDay provides simple integration of third-party algorithms into the machine learning pipeline enabling it to adapt to new missions. GameDay runs on the Kubernetes platform making it easy to deploy it in fixed sites, the cloud, or the edge. GameDay reduces the big-data problem inherent in persistent ISR collection and latency in manual exploitation.