Depth based dual component dynamic gesture recognition

Helman Stern, Kiril Smilansky, Sigal Berman

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Scopus citations

Abstract

In this paper we describe several approaches for recognition of gestures that include simultaneous arm motion and hand configuration variations. Based on compound (dual component) gestures selected from the ASL we developed methods for recognizing such gestures from Kinect sensor videos. The method consists of hand segmentation from depth images followed by feature extraction based on block partitioning of the hand image. When combined with trajectory features a single-stage classifier is obtained. A second method which classifies arm movement and hand configuration in two-stages is also developed. These two methods are compared to a moment based classifier from the literature. Using a database of 11 subjects for training and testing the average classification accuracy of the one-stage classifier was the highest (95.5%) and that of the moment based classifier was the lowest (20.9%). The two-stage classifier obtained an average classification accuracy of61.1%.

Original languageEnglish
Title of host publicationProceedings of the 2013 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2013
EditorsHamid R. Arabnia, Leonidas Deligiannidis, Joan Lu, Fernando G. Tinetti, Jane You, George Jandieri, Gerald Schaefer, Ashu M. G. Solo, Vladimir Volkov
PublisherCSREA Press
Pages985-991
Number of pages7
Volume2
ISBN (Electronic)1601322534, 9781601322531
StatePublished - 1 Jan 2013
Event2013 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2013, at WORLDCOMP 2013 - Las Vegas, United States
Duration: 22 Jul 201325 Jul 2013

Conference

Conference2013 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2013, at WORLDCOMP 2013
Country/TerritoryUnited States
CityLas Vegas
Period22/07/1325/07/13

Keywords

  • Dynamic motion gestures
  • Gesture recognition
  • Human-machine interaction
  • Sign language

ASJC Scopus subject areas

  • Computer Graphics and Computer-Aided Design
  • Computer Vision and Pattern Recognition

Fingerprint

Dive into the research topics of 'Depth based dual component dynamic gesture recognition'. Together they form a unique fingerprint.

Cite this