| Title |
Swing Sequence Images based 3D Golf Swing Analysis using Deep Learning |
| Authors |
고경리(Kyeong-Ri Ko) ; 반성범(Sung Bum Pan) |
| DOI |
https://doi.org/10.5573/ieie.2019.56.10.49 |
| Keywords |
golf swing analysis; sequence image regression; deep learning |
| Abstract |
The evaluation and analysis of the fast moving golf swing motion by human eyes can vary greatly depending on the evaluator's perspective. In this paper, we study the method of three-dimensional quantitative information extraction by overcoming the limitation of single camera based golf swing analysis system using deep learning which is showing good performance in image recognition field recently. First, the features of the sequence images are extracted using convolutional neural network, and the swing section is classified. Sequence features with swing section information are used as inputs to the bidirectional long short-term memory based swing analysis model, and perform body-swing, head-up, and X-factor value prediction. Experimental results showed that the performance of the upper body motion prediction RMSE 4.23, the head motion prediction RMSE 5.18, and the X-factor prediction result RMSE 3.86. As a result, it is confirmed that 2D frontal sequence images based 3D quantitative golf swing analysis is possible. |