Classification techniques for smartphone based activity detection

Research output: Contribution to conferencePaperpeer-review

Abstract

In a world where the lack of physical activity is becoming alarmingly prevalent, the accurate recognition of human movement, or the lack thereof, has never been more important. In this paper, the authors describe a method of accurately detecting human activity using a smartphone accelerometer paired with a dedicated chest sensor. A series of trials were carried out in Ireland, initially involving N = 6 individuals to test the feasibility of the system, before a final trial with N = 24 subjects took place in the Netherlands. The protocol used and analysis of some 1400 minutes of recorded datum from this latter trial are described in detail throughout this paper. The design, implementation, testing and validation of a custom mobility classifier is also discussed. Offline analysis using machine learning algorithms, including C4.5, CART, SVM, Multi-Layer Perceptrons, and Naïve Bayes was carried out. Methods were also deployed which allow existing fixed position based algorithms to function in an orientation independent manner. Analysis of collected datum indicate that accelerometers placed in these locations, are capable of recognising activities including sitting, standing, lying, walking, running and cycling with accuracies as high as 98%.

Original languageEnglish
Pages154-158
Number of pages5
DOIs
Publication statusPublished - 2012
Event2012 11th IEEE International Conference on Cybernetic Intelligent Systems, CIS 2012 - Limerick, Ireland
Duration: 23 Aug 201224 Aug 2012

Conference

Conference2012 11th IEEE International Conference on Cybernetic Intelligent Systems, CIS 2012
Country/TerritoryIreland
CityLimerick
Period23/08/1224/08/12

Keywords

  • Accelerometers
  • Ambient Assisted Living
  • Machine Learning
  • Physical Activity Classification

Fingerprint

Dive into the research topics of 'Classification techniques for smartphone based activity detection'. Together they form a unique fingerprint.

Cite this