We offer a groundbreaking technical perspective for table tennis players, coaches and fans. The features of our solution includes object identification and localization, semantics detection, and actions and tactics recognition. Different playing styles can be identified by tracking 22 key points of the players bodies. Furthermore, we are able to track players, ball and racket in real-time. With high frequency and high resolution cameras, a 99% accuracy is achieved in real-time ball tracking. For real-time player tracking, player bounding boxes are identified and tracked which allows innovative KPIs.