Fig. 3From: Robust head pose estimation based on key frames for human-machine interactionKey frame learning process. Each figure represents the same target observed from three views. The discretized orientation is depicted as the region of the sphere. The color indicates visited areas (green) from regions without an associated pose (yellow). The set of points in red is the 3D model of the estimated pose. Target orientation at current frame is shown with the blue lineBack to article page