Detection and recognition of human body posture in motion based on sensor technology

Abstract

The recognition of human motion posture is of great values in the field of sports. In this paper, inertial sensor technology is employed to recognize four postures including dribbling, passing, catching, and shooting in basketball. The data are collected by four nine‐axis inertial sensors worn on the arm. The time domain and frequency domain features are extracted after the process of smoothing and normalizing. Then 30 eigenvectors are obtained by dimensionality reduction of principal component analysis (PCA), and the support vector machine (SVM) method is used for posture recognition. The experimental results show that after PCA dimensionality reduction, the recognition accuracy of the features as the SVM input is significantly high. Compared with the recognition accuracy of back propagation neural network (BPNN) (85.4%), the average recognition accuracy of SVM is 96%, which verifies the reliability of the method proposed in this study. This research proves the effectiveness of sensor technology in basketball posture recognition, which can provide a reliable guidance for basketball training. © 2020 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

The Gaitprint: Identifying Individuals by Their Running Style

“…They have also been shown to be convenient and easy-to-use in different practical sport settings and rehabilitation related contexts [16,17]. A recent study [18] in the sport of basketball focused on the use of MEMS sensors to detect complex, sports-typical movements of the upper extremities, such as dribbling, catching, passing, and throwing. The group used well-known computer-based evaluation methods (Principal component analysis and support vector machine) and presented a highly reliable method with an average recognition rate of 96%.…”

Motion Recognition Based on Sum of the Squared Errors Distribution

“…As shown in the Table Ⅸ, WTS has higher segmentation rate and more stable segmentation effect. We compare the recognition rate of BPTS with that of other algorithms, such as Average Threshold Crossing (ATC) [77], CNN [78], Deep Forest algorithm (DF) [79], Deep Convolutional Network (DCN) [80], Dynamic Time Warping (DTW) [81], PCA [82] and SVM [83], WPT and unscented Kalman neural network (UKFNN) [84], Hybrid Bidirectional Unidirectional Long Short-Term Memory (HBU-LSTM) [85], and PCA and SVM [13]. Some parameters or structures of other algorithms are configured as shown in Table Ⅹ.…”

“…Although Euler angles have the phenomenon of gimbal lock, the movement space of the human arm itself has some limitations, and it is difficult to trigger the gimbal lock within its movement range. (10)- (13) are used to convert the quaternion into Euler angles.…”

“…Therefore, we extract the feature parameters of simple motions, which only contain location information. The location-related feature values are as follows: RMS (root mean square), AVG (average), IQR (interquartile range), MAD (mean absolute deviation), VAR (variance value), SD (standard deviation), MAX (maximum value), and MIN (minimum value) [13], [69], [70]. RMS is used to measure the deviation between the observed value and the true value, and it is used to measure the relationship between the measured value and the true value, which is not applicable in this paper.…”

“…However, there are some limitations, such as the camera's field of view, the complexity of target background, the intensity of light, and the privacy of the user. With the development of sensor technology, non-visual motion recognition technology has been developed rapidly, such as Hidden Markov Model (HMM) [12], Support Vector Machine (SVM) [13], Principal component analysis (PCA) [14], K-Nearest Neighbor (KNN) [15], Neural Network (NN), Singular value decomposition (SVD), and Wavelet packet transform (WPT). Most of these algorithms use surface electromyography (SEMG) signal acquisition sensor and inertial sensors in the process of data collection.…”