The 21st-century target recognition is here:
It’s called DOCTRINAIRE.
DOCTRINAIRE combines the reliability of tried-and-true target detection doctrine with the power of modern artificial intelligence and machine learning approaches. The software is explainable, robust, and—best of all—can be configured in the field to detect new kinds of targets without network re-training.
How DOCTRINAIRE Works
Traditionally, training doctrine has taught warfighters to detect and classify vehicles using the fundamental building blocks, i.e., hull, armaments, turret, suspension/wheels (HATS) or to detect aircraft by identifying wings, engine, fuselage, tail (WEFT).
By decomposing a detected vehicle into its component parts, DOCTRINAIRE offers much more capability. It can estimate not only the target class, but also the target pose, part articulations, and target range.
DOCTRINAIRE is robust because it does not require training data for all possible vehicles in all environments. It is explainable because target class declarations are explicitly based on target detection doctrine. What’s more, with DOCTRINAIRE, systems can be re-configured in the field because new target classes can be identified with only a CAD model of the target and no additional re-training.
See how DOCTRINAIRE is changing target recognition for the Department of Defense and intelligence community.
Start building your customized DOCTRINAIRE solution with us today.