Developing an Intelligent Table Tennis Umpiring System
Researcher: Dr. Patrick Wong
Table tennis is a fast sport. A service usually takes a few seconds to complete but an umpire needs to make up to 31 observations and makes a judgment before or soon after the service is complete. This is a complex task so adopting computerised tools to provide fast measurement information for umpires will help considerably in aiding their judgments. The aim of this research is to develop an intelligent system which is able to identify and track the location of the ball from live video images and evaluate the service according to the service rules. Techniques of identifying a table tennis ball from video sequences have been developed. A number of specialised algorithms have been employed to identify and measure the characteristics of the ball. Artificial neural networks have been applied as a classifier. It classifies whether the detected object is not-a- ball, a ball on the palm or a ball in mid air. The system has been tested on a large number of still images and video sequences which contain distinctive characteristics in terms of the type of motion and object occlusion. The preliminary results are very promising. To obtain more accurate and reliable measurements, use of multiple cameras are proposed. The cameras are situated at positions where the umpire and assistant umpire are. Additional cameras may be fixed high above the table to take aerial views. With many data now feeding in from these cameras, it is suggested to employ a Multi-agent System to coordinate these complicated processes. Each camera is associated with an agent who can make independent decision based on the data it received. These local decisions can then be fed in to a higher hierarchy agent for consideration of the final decision.
Some experiment results can be found here. Publications of this work can be found at Open Research Online.