Robust eye center localization based on an improved SVR method

Zhiyong Wang, Haibin Cai, Honghai Liu*

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

99 Downloads (Pure)

Abstract

Eye center localization is an important technique in gaze estimation, human computer interaction, virtual reality, etc., which attracts a lot of attention. Although a great deal of progress has been achieved over the past few years, the accuracy declines dramatically due to the low input image resolution, poor lighting conditions, side face, and eyes status such as closed or covered. To handle this issue, this paper proposes an improved support vector regression (SVR) method to detect the eye center based on the facial feature localization. Several image processing techniques were tried to improve the accuracy, and results showed that the SVR combining a Gaussian filter could get a better accuracy.

Original languageEnglish
Title of host publicationNeural Information Processing - 25th International Conference, ICONIP 2018, Proceedings
EditorsSeiichi Ozawa, Andrew Chi Sing Leung, Long Cheng
PublisherSpringer Verlag
Pages623-634
Number of pages12
ISBN (Electronic)978-3-030-04239-4
ISBN (Print)978-3-030-04238-7
DOIs
Publication statusPublished - 18 Nov 2018
Event25th International Conference on Neural Information Processing - Siem Reap, Cambodia
Duration: 13 Dec 201816 Dec 2018

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume11307
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference25th International Conference on Neural Information Processing
Abbreviated titleICONIP 2018
Country/TerritoryCambodia
CitySiem Reap
Period13/12/1816/12/18

Keywords

  • Eye center localization
  • Gaussian filter
  • SVR

Fingerprint

Dive into the research topics of 'Robust eye center localization based on an improved SVR method'. Together they form a unique fingerprint.

Cite this