Improving Supervised Classification of Activities of Daily Living Using Prior Knowledge

Constant growing of the mean age of the population and bottleneck created at the entry of institutions makes telemedicine for elderly people an actual challenge largely explored. It requires recognizing the behavior and actions of a person inside his home with non-intrusive sensors and to process data to check his evolution. This paper presents the results of the study of prior introduction, in Support Vector Machine, to improve this automatic recognition of Activities of Daily Living. From a set of activity performed in a smart home in Grenoble, we obtained models for seven activities of Daily Living and test the performances of this classification and the introduction of spatial and temporal priors. Finally, we discuss the different results.