Piškot (Cookie) is a seminar series organized by the Faculty of Computer and Information Science at the University of Ljubljana. This time the topic will be activity monitoring with machine learning using smartphone sensor data. The speaker will be SLAIS member Mitja Luštrek.
A smartphone can easily tell how the user moves about with the help of the GPS, but it can also recognize elementary activities such as walking, running, sitting etc. quite accurately. Furthermore, it can estimate the calories the user burns in a more sophisticated fashion than the widely available step counter applications. This requires laborious collection and labelling of sensor data, using machine learning to build several models for different phone placements, and assembling them in an activity-analysis pipeline. In this Piškot we will take a look at such a pipeline and demonstrate a mobile application that uses it.