Text this: Actor–Critic Reinforcement Learning and Application in Developing Computer-Vision-Based Interface Tracking