Personalization algorithm for real-time activity recognition using PDA, wireless motion bands, and binary decision tree

Authors:
Juha Pärkkä;Luc Cluitmans;Miikka Ermes
Affiliations:
VTT Technical Research Centre of Finland, Tampere, Finland;VTT Technical Research Centre of Finland, Tampere, Finland;VTT Technical Research Centre of Finland, Tampere, Finland
Venue:
IEEE Transactions on Information Technology in Biomedicine
Year:
2010

Citing 5
Cited 4

Practical characteristics of neural network and conventional pattern classifiers on artificial and speech problems

Advances in neural information processing systems 2
Pattern Classification (2nd Edition)

Pattern Classification (2nd Edition)
Automatic feature selection for context recognition in mobile devices

Pervasive and Mobile Computing
Activity classification using realistic data from wearable sensors

IEEE Transactions on Information Technology in Biomedicine
Detection of Daily Activities and Sports With Wearable Sensors in Controlled and Uncontrolled Conditions

IEEE Transactions on Information Technology in Biomedicine

Rejection of irrelevant human actions in real-time hidden Markov model based recognition systems for wearable computers

Proceedings of the 2nd Conference on Wireless Health
Towards global aerobic activity monitoring

Proceedings of the 4th International Conference on PErvasive Technologies Related to Assistive Environments
Creating and benchmarking a new dataset for physical activity monitoring

Proceedings of the 5th International Conference on PErvasive Technologies Related to Assistive Environments
Personalized mobile physical activity recognition

Proceedings of the 2013 International Symposium on Wearable Computers

Quantified Score

Hi-index	0.00

Visualization

Abstract

Inactive and sedentary lifestyle is a major problem in many industrialized countries today. Automatic recognition of type of physical activity can be used to show the user the distribution of his daily activities and to motivate him into more active lifestyle. In this study, an automatic activity-recognition system consisting of wireless motion bands and a PDA is evaluated. The system classifies raw sensor data into activity types online. It uses a decision tree classifier, which has low computational cost and low battery consumption. The classifier parameters can be personalized online by performing a short bout of an activity and by telling the system which activity is being performed. Data were collected with seven volunteers during five everyday activities: lying, sitting/standing, walking, running, and cycling. The online system can detect these activities with overall 86.6% accuracy and with 94.0% accuracy after classifier personalization.