IMU-Based walking workouts recognition

Fanuel Wahjudi, Fuchun Joseph Lin

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

7 Scopus citations

Abstract

To better accurately estimate the calories burnt during popular walking workouts, it is essential to detect the environment under which these workouts are conducted. To our best knowledge, no gait analysis studies have been done so far for such detection. This research addresses this problem by recognizing walking workouts under different environments based on the foot-mounted inertial sensor. Our objective is to recognize ten different workout activities including walking and brisk-walking under flat surface, ascending/descending staircase and upward/downward slope with no stairs. Our algorithm first identifies the extended foot-flat phase, then uses it as a boundary to extract key important features. Decision Tree, Random Forest and K-Nearest Neighbor machine learning algorithms are evaluated to decide which one works the best along with our algorithm.

Original languageEnglish
Title of host publicationIEEE 5th World Forum on Internet of Things, WF-IoT 2019 - Conference Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages251-256
Number of pages6
ISBN (Electronic)9781538649800
DOIs
StatePublished - Apr 2019
Event5th IEEE World Forum on Internet of Things, WF-IoT 2019 - Limerick, Ireland
Duration: 15 Apr 201918 Apr 2019

Publication series

NameIEEE 5th World Forum on Internet of Things, WF-IoT 2019 - Conference Proceedings

Conference

Conference5th IEEE World Forum on Internet of Things, WF-IoT 2019
Country/TerritoryIreland
CityLimerick
Period15/04/1918/04/19

Keywords

  • activity recognition
  • environment detection
  • Gait analysis
  • machine learning algorithms
  • walking workouts

Fingerprint

Dive into the research topics of 'IMU-Based walking workouts recognition'. Together they form a unique fingerprint.

Cite this