Golf Swing and Human Motion Estimation Using an RGB-D Camera and a Club-Mounted IMU

Abstract

This article presents a unified framework that integrates an external red–green–blue–depth (RGB-D) camera with a club-mounted inertial measurement unit (IMU) to jointly estimate golf club motion and full-body golfer motion. The proposed approach employs an indirect Kalman filter to fuse IMU measurements with visual observations of the golf club, enabling robust estimation of club motion while compensating for sensor drift, calibration errors, and mounting uncertainties. The estimated club motion is further integrated with 3-D joint observations within an optimization-based skinned multi-person linear model-expressive (SMPL-X) framework for human motion estimation, where both 3-D joint constraints and hand–club interaction constraints enforce physical consistency between the golfer and the club. Experimental results demonstrate that the proposed method significantly improves clubhead trajectory estimation compared to IMU-only approaches, achieving a mean position error of 2.8 cm. In addition, the method produces coherent and physically plausible full-body motion reconstruction. These results highlight the effectiveness of the proposed system for comprehensive and reliable golf swing analysis.