Deep belief network based gaze tracker for auto-aiming system

This paper proposes a design of an auto-aiming system, which is capable of tracking a user's gaze automatically. The auto-aiming system is composed of an electric gun control system and a wearable gaze tracker. According to a user's gaze direction which is captured by the helmet camera, the gun can automatically aim an object which the user is looking at. The mapping model between the user's eye movement and the shooting direction of the gun is established by a Deep Belief Network (DBN). This DBN-based gaze tracking model allows the auto-aiming shooting system to aim the target timely and accurately. Finally, the experiment results demonstrate the proposed auto-aiming system for different users can achieve an average accuracy of 96%.