Model-Based Motion Tracking of Infants

Mikkel Damgaard Olsen, Anna Herskind, Jens Bo Nielsen, Rasmus Reinhold Paulsen

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

9 Downloads (Pure)

Abstract

Even though motion tracking is a widely used technique to analyze and measure human movements, only a few studies focus on motion tracking of infants. In recent years, a number of studies have emerged focusing on analyzing the motion pattern of infants, using computer vision. Most of these studies are based on 2D images, but few are based on 3D information. In this paper, we present a model-based approach for tracking infants in 3D. The study extends a novel study on graph-based motion tracking of infants and we show that the extension improves the tracking results. A 3D model is constructed that resembles the body surface of an infant, where the model is based on simple geometric shapes and a hierarchical skeleton model.
Original languageEnglish
Title of host publicationComputer Vision – ECCV 2014 Workshops : Proceedings, Part III
EditorsLourdes Agapito, Michael M. Bronstein, Carsten Rother
PublisherSpringer
Publication date2014
Pages673-685
ISBN (Print)978-3-319-16198-3
ISBN (Electronic)978-3-319-16199-0
DOIs
Publication statusPublished - 2014
Event2nd Workshop on Assistive Computer Vision and Robotics (ACVR 2014) - Zürich, Switzerland
Duration: 12 Sep 2014 → …
Conference number: 2
http://www.ino.it/ACVR2014/

Workshop

Workshop2nd Workshop on Assistive Computer Vision and Robotics (ACVR 2014)
Number2
CountrySwitzerland
CityZürich
Period12/09/2014 → …
OtherIn conjunction with European Conference on Computer Vision (ECCV'14), September 6-12, 2014
Internet address
SeriesLecture Notes in Computer Science
Volume8927
ISSN0302-9743

Keywords

  • 3D Model Fitting
  • Infant Pose Estimation
  • Markerless Motion Tracking
  • Depth Images

Fingerprint Dive into the research topics of 'Model-Based Motion Tracking of Infants'. Together they form a unique fingerprint.

Cite this