Real-time visual tracking by deep reinforced decision making

This paper introduces a novel real-time visual tracking algorithm based on a template selection strategy constructed by deep reinforcement learning methods that is generally applicable to other confidence map based tracking algorithms.