motivation
Activity recognition plays a vital role enabling the vision of ubiquitous computing. For computing to be pro-active and help us during real-time, real-life activities, computing needs to aware of us and the current environment we are in, so to say, our context. Most of today's context and activity recognition systems use probabilistic machine learning algorithms that need extensive training data with exact activity labels lto work. In addition, most algorithms assume fixed set of sensors with known location and orientation. Therefore, context and activity recognition, although widely used in lab settings, did not really make it into our everyday life. This webpage focuses on efforts to enable researchers and developers a easier use of activity recognition technology in their projects. For more background information go here or check out the publications/talks section.In the following, you can see some pictures of experiments done from our lab. You will agree with me, 'normal' people don't really want to dress like that. The sensor placement is fine for some data collection, yet to use this technology in everyday life will still take some more effort ;)
