Nowadays several types of sensors exist in the industry and markerless sensors are considered the cheapest. However, if used correctly, these sensors can substitute expensive ones when measuring movement data. As these sensors do not use marker technology, the data stream can consist of measurement errors even if the user is in a stationary state.

The goal is to create a filter for a real-time data stream. The aim of this filter is to correct the data stream and to allow for a better percentage of gesture recognition.