NCTU

Golf Swing Training System

2017/06/09
CodeM15-HTC-14-1
ProfessorHua-Tsung Chen
Applicationthe applications of education
FunctionAR (Augmented reality) and Image detection
Technical BenefitCost-reduction and Easy operation/ reduce participation of operators
Technology StatusCan be transferred

Self-training is essential in sports exercise. However, without the instruction of a coach, a practitioner may progress to a limited extent. Improper postures may even cause serious harm to muscles and ligaments of the body. Hence, we utilize computer vision technology to analyze the swing posture of a golfer, compute the body contour and skeleton, and then generate feature lines of the head, hips, and the spine. Involved with golf training knowledge, some reference lines can be generated automatically to judge whether the swing posture is correct. Visualized and auditory instructions are also provided, informing the user to adjust his/her posture. In addition, we have also implemented this system with intelligent glasses. The detected feature lines and visualized instructions can be transmitted to the intelligent glasses, enabling the user to rectify his/her posture immediately. The experiments show that under a normal lighting condition, the accuracy of feature line detection can be higher than 90% for the head and hips, and can be higher than 85% for the spine. As for the computational efficiency, this system can run in real time for the image resolutions of 640´360 and 960´540 (the processing speeds are about 55 fps and 30 fps, respectively).

(a)

(b)

Fig. (a) Architecture of the prposed golf swing training system. (b) Detection of the head, hips, and spine feature lines during a swing.

Demo video