Head direction estimation from low resolution images with scene adaptation

• Published in: Computer vision and image understanding Vol. 117; no. 10; pp. 1502 - 1511
• Main Authors: Chamveha, Isarun, Sugano, Yusuke, Sugimura, Daisuke, Siriteerakul, Teera, Okabe, Takahiro, Sato, Yoichi, Sugimoto, Akihiro
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier Inc 01-10-2013; Elsevier
• Subjects: Adaptation; Aggregates; Appearance-based approach; Applied sciences; Artificial intelligence; Computer science; control theory; systems; Computer vision; Exact sciences and technology; Graph-based image segmentation; Head direction estimation; Image acquisition; Labels; Low resolution image; Pattern recognition. Digital image processing. Computational geometry; Scene adaptation; Segments; Similarity; Unsupervised learning; Graph-based image segmentation; Appearance-based approach; Head direction estimation; Scene adaptation; Unsupervised learning; Low resolution image; Legged locomotion; Head; Image resolution; Image processing; Graph theory; Low resolution; Ground truth; Walking; Image segmentation; Experimental result; Image sequence; Scene analysis
• ISSN: 1077-3142; 1090-235X
• DOI: 10.1016/j.cviu.2013.06.005

•A head pose estimation method that adapts to individual scenes is proposed.•The method automatically collects training dataset for head pose estimators.•The method handle appearance differences within the same scene.•The method demonstrate high performance without manually collected training data.
This paper presents an appearance-based method for estimating head direction that automatically adapts to individual scenes. Appearance-based estimation methods usually require a ground-truth dataset taken from a scene that is similar to test video sequences. However, it is almost impossible to acquire many manually labeled head images for each scene. We introduce an approach that automatically aggregates labeled head images by inferring head direction labels from walking direction. Furthermore, in order to deal with large variations that occur in head appearance even within the same scene, we introduce an approach that segments a scene into multiple regions according to the similarity of head appearances. Experimental results demonstrate that our proposed method achieved higher accuracy in head direction estimation than conventional approaches that use a scene-independent generic dataset.