Appearance-Based Head Pose Estimation with Scene-Specific Adaptation

We propose an appearance-based head pose estimation method that can be automatically adapted to individual scenes. Appearance-based estimation methods usually re-quire a ground-truth dataset taken from a scene which is similar to test video sequences. However, it is almost im-possible to acquire many manually-labeled head images for each scene. To address the problem, we introduce a new ap-proach for aggregating ground truth head pose labels auto-matically by inferring head pose labels from walking direc-tion. Experimental results demonstrate that our proposed method achieves better accuracy in head pose estimation than the conventional approach using a scene-independent generic dataset. 1