University of Twente Student Theses
Activity classification using a hip worn inertial sensor
Pilagkas, Christos (2012) Activity classification using a hip worn inertial sensor.
Full text not available from this repository.
| Full Text Status: | Access to this publication is restricted |
| Abstract: | A telemedicine project is under development for energy expenditure (EE) estimation. This project concerns the assessment of estimating EE by implementing and testing a classification algorithm for recognizing five different sedentary and dynamic daily human activities – lying, sitting, standing, walking and cycling. During the experiments 4 inertial sensors were used in various positions and orientations around the waist of each one of 3 different healthy subjects. The classification algorithm is based on the SVM machine learning technique using a 14D time domain feature space extracted from fixed window size. Two different fall-back strategies of the initially chosen methods were used. During the first, the classification algorithm turned to orientation dependent by adding the gravitational acceleration signal to the features. The second alternative was concerning the windowing technique, where an activity defined windowing technique was neglected and a fixed window size was implemented. Finally, five different window sizes, two different feature extraction methods, three different cross validations, and two different alignment techniques for the predicted labels were used during the whole project. The results showed a best accuracy of 92.2% ± 2.5% for the 4 sec. window size and about 91.4% ± 1.4% for the 6 sec. window size for detecting the sequence of the activities using leave-one-subject-out cross validation technique performed over different subjects. Higher accuracy can be obtained using a sensor at a semi-fixed position near the hip area and orientation dependent features |
| Item Type: | Internship Report (Master) |
| Faculty: | ET: Engineering Technology |
| Subject: | 52 mechanical engineering |
| Programme: | Biomedical Engineering MSc (66226) |
| Link to this item: | https://purl.utwente.nl/essays/63962 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page