Range image sensor based eye gaze estimation by using the relationship between the face and eye directions

Haruya Tamaki, Tsugunosuke Sakai, Yosuke Ota, Fusako Kusunoki, Shigenori Inagaki, Ryohei Egusa, Masanori Sugimoto, Hiroshi Mizoguchi

Research output: Contribution to journalArticle

Abstract

In this paper, we estimate the eye gaze by using the face direction to estimate the object of interest in an extensive area such as an educational display or museum exhibit. However, the direction of the gaze may differ from that of the face. Therefore, our approach was to use a method that utilizes both the face and eye directions to improve the accuracy of our estimation compared to only using the face direction. The first part of our study involved apprehending the relationship between the direction of the face and the eyes. The second part entailed estimating the eye gaze using this relationship. We then evaluated the effectiveness of this method in comparison to using the relationship with the face direction only.

Original languageEnglish
Pages (from-to)2297-2308
Number of pages12
JournalInternational Journal on Smart Sensing and Intelligent Systems
Volume9
Issue number4
Publication statusPublished - 1 Jan 2016

    Fingerprint

Keywords

  • Active appearance models
  • Eye gaze direction
  • Face direction
  • Range image sensor
  • Relationship between face and eyes

Cite this