Application of radar-based technologies in real-world scenarios

Abstract

The privacy issue is an important topic nowadays. With that, some real-world application requires working with such kind of data. For example, surveillance, or prayer tracking and assistance system. In this thesis, we develop a framework based on radar technologies, that can satisfy the privacy requirements, and provide enough capabilities to meet the system's demands.

The usage of radar for some kind of surveillance is on the rise, but there is no research done for prayer tracking and assistance. Also, there is no full descriptive work that covered all aspects of the implementation of such a system.

This thesis provides a tool to develop a prayer tracking and assistance system. This approach can be used for a broad variety of scenarios. The main pillar of the system is the classification algorithm. We developed and compared different algorithms and found that convolutional neural networks provide the best results with accuracy on completely unknown data up to 98.2 percent.