Teaching and Research Section of Physical Education, College of Sport, Neijiang Normal University, Neijiang 641112, China.
Centre for Sport and Exercise Sciences, University Malaya, Kuala Lumpur 50603, Malaysia.
Int J Environ Res Public Health. 2022 Feb 3;19(3):1744. doi: 10.3390/ijerph19031744.
This study aimed to assess the motion accuracy of Baduanjin and recognise the motions of Baduanjin based on sequence-based methods. Motion data of Baduanjin were measured by the inertial sensor measurement system (IMU). Fifty-four participants were recruited to capture motion data. Based on the motion data, various sequence-based methods, namely dynamic time warping (DTW) combined with classifiers, hidden Markov model (HMM), and recurrent neural networks (RNNs), were applied to assess motion accuracy and recognise the motions of Baduanjin. To assess motion accuracy, the scores for motion accuracies from teachers were used as the standard to train the models on the different sequence-based methods. The effectiveness of Baduanjin motion recognition with different sequence-based methods was verified. Among the methods, DTW + -NN had the highest average accuracy (83.03%) and shortest average processing time (3.810 s) during assessing. In terms of motion reorganisation, three methods (DTW + -NN, DTW + SVM, and HMM) had the highest accuracies (over 99%), which were not significantly different from each other. However, the processing time of DTW + -NN was the shortest (3.823 s) compared to the other two methods. The results show that the motions of Baduanjin could be recognised, and the accuracy can be assessed through an appropriate sequence-based method with the motion data captured by IMU.
本研究旨在评估八段锦的动作准确性，并基于序列方法识别八段锦的动作。通过惯性传感器测量系统 (IMU) 测量八段锦的运动数据。招募了 54 名参与者来捕捉运动数据。基于运动数据，应用了各种基于序列的方法，即动态时间规整 (DTW) 结合分类器、隐马尔可夫模型 (HMM) 和递归神经网络 (RNN)，以评估动作准确性并识别八段锦的动作。为了评估动作准确性，使用教师的动作准确性分数作为标准，在不同的基于序列的方法上训练模型。验证了不同基于序列的方法识别八段锦运动的有效性。在这些方法中，DTW+-NN 在评估过程中的平均准确率最高（83.03%），平均处理时间最短（3.810 秒）。在动作重组方面，三种方法（DTW+-NN、DTW+SVM 和 HMM）的准确率最高（均超过 99%），彼此之间没有显著差异。然而，与其他两种方法相比，DTW+-NN 的处理时间最短（3.823 秒）。结果表明，通过 IMU 捕获的运动数据，可以使用适当的基于序列的方法识别八段锦的动作，并且可以评估其准确性。
