The U.S. Army has identified a need to identify an unknown subject in those circumstances and environments in which it is only possible to collect a profile face image (i.e., images extracted from Captured Enemy Material (CEM) and from publicly available information sources). The ability to identify individuals in currently fielded systems based on a profile face image would be greatly enhanced with a face matching algorithm optimized for profile matching. The Project Manager (PM) is interested in algorithms that incorporate state-of-art artificial intelligence (AI) and machine learning (ML) processes using modern neural network architectures that are designed to perform face matching of profile images. To address this need, Tygart Technology proposes to use our experience developing and deploying ML models to investigate and document current state-of-the-art techniques for face recognition and their applicability to profile matching, investigate the feasibility and approach for developing a profile face recognition algorithm, and identify the protocol for employment of the resulting capability with current identity operations and intel platforms. In the Phase I â Option, Tygart will validate the model design, overall architecture and model training pipeline by developing a working ML model capable of performing profile face recognition
