In the following we are going to present the working principle of two computer vision concepts, which are of great value for sports science: Tracking and human pose estimation. We are going to present these complex methods and their underlying working principles. We also provide the tools, that sport scientists are able select the most effective algorithm or implementation for their specific case. Additionally, we give real world examples of how to make use of these concepts for sport science and address the most common challenges. As these algorithms often were engineered for everyday tasks the adaption to the domain specific tasks is of high importance. As this process often requires the collection of sports data, we show the most important points that need to be considered.