Weapon operators and analysts use automated systems and prior knowledge to plan and execute missions involving target detection and tracking. When automated systems operate with a high degree of tracking accuracy, they provide immense value to analysts and operators and improve their efficiency. Latest advances in deep learning algorithms and improvements in computational hardware have led to significant improvements in target detection and tracking capabilities. However, human intervention is often required to correct, provide additional information and improve tracking performance. In this effort, we propose to develop a joint human-machine teaming approach for tracking. The tracker will learn from its failures over time and learn to anticipate failures. Tracking accuracy and user intervention time will be used to quantify the performance of the proposed effort.
