Publication at Faculty of Mathematics and Physics |
2011
Abstract
We present our initial work on an "automatic diary" recognizing and storing episodes of human daily activities. The goal is to create an application that will perform activity recognition based on data collected from a mobile phone.
This includes a GPS location, WiFi and Bluetooth signals. Our aim is to combine these sensory data with information from publicly available databases of points of interest thus identifying restaurants, schools etc.
Until now we have collected several months of hierarchically annotated data, created a desktop viewer of the logged data and experimented with inference of activities on two different levels of abstraction. Several machine learning algorithms were tested in these experiments.