Visual Golf Club Tracking for Enhanced Swing Analysis

This paper presents a new visual tracking technology that relies on the use of a global motion model to achieve robustness. We demonstrate its effectiveness for the purpose of retrieving the 2D spatio-temporal trajectory of a golf club head from ordinary video sequences of golf swings, so that information about club orientation, local speed and acceleration can also be obtained. We have integrated it into a fully automated system that requires neither user intervention nor the use of instrumented golf club or clothing, and that is usable in a natural environment with a potentially cluttered background. Our algorithm robustly fits a global swing trajectory model to club location hypotheses obtained from single frames. This process makes our approach very robust and it will soon be integrated into a commercial product. Several experimental results are presented to illustrate the success of this new method.