This letter presents a novel estimator and predictor framework for target tracking applications that estimates the pose of a mobile target that intermittently leaves the field-of-view (FOV) of a mobile agent’s camera. Specifically, the framework uses an attention deep motion model network (DMMN) to estimate the dynamics of the target when the target is in the agent’s FOV and uses the DMMN to predict the position, orientation, and velocity of the target when the target is outside the agent’s FOV. A Lyapunov-based stability analysis is performed to determine the maximum dwell-time condition on target measurement availability, and experimental results are provided to demonstrate the performance of the proposed framework.