In the paper, we describe technical details of multi-player sports tracker system. We demonstrate that proper in-depth analysis of video frames  sequence may provide a lot of useful information required for sports  analytics. Object detection and tracking steps are analyzed. Novel ideas for efficient filtering of false detections and irrelevant tracks are proposed.  Also, we show an example of how the object tracking information used for  the regions of interest location allows streaming sports events without  human operators. At last, important practical implementation details, as  well as initial experimental results, are discussed.

Light-Weight Tracker for Sports Applications