The multidisciplinary area of GEOINT is changing and becoming more complex. A major driver of innovation in GEOINT collection and processing is artificial intelligence (AI). AI is being leveraged to help accomplish spatial analysis, change detection, and image or video triage tasks where filtering objects of interest from large volumes of data is critical. NGA is confronting the changing GEOINT landscape with emerging data models and development campaigns such as the 2020 Analysis Technology Plan and Strategy 2025. In response to this, we have developed methods to rapidly build datasets consisting of large volumes of physically realistic infrared ground order of battle exemplar images. We have shown that these datasets can be used to train and test machine learning algorithms for recognition of real-world targets. To further this work, we propose to build a software prototype that can train machine learning algorithms for defense applications by the following: i) procedural generation of large volumes of pertinent infrared imagery, and ii) state-of-the-art machine learning and explainable AI for interpretable target recognition and feature extraction. We will validate this prototype with a culminating experiment to show that radiometrically-accurate synthetic data can be used to train machine learning algorithms for predictable real-world performance.
