ArticlePublisher preview available

SVM-Adaboost based badminton offensive movement parsing technique

Authors:
Chun-Yao Shih
Chun-Yao Shih
Chun-Yao Shih
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Yong-Tao Lin
Yong-Tao Lin
Yong-Tao Lin
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Wei Chen
Wei Chen
Wei Chen
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Jui-Chan Huang
Jui-Chan Huang
Jui-Chan Huang
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Read publisher preview
Download citation
To read the full-text of this research, you can request a copy directly from the authors.

Abstract and Figures

With the development of smart devices and network technology, sports data statistics systems such as smart bracelets have also been widely studied, but there are fewer studies on racket sports data analysis systems such as badminton and other sports with complex movements, so this study will take badminton as the object of study, choose the inertial sensor-based data acquisition method to collect real-time data, choose a combination of sliding window and action window to carry out data segmentation, and use the principal component analysis is used to achieve data dimensionality reduction, and finally a two-layer classification algorithm based on support vector machine and adaptive boosting algorithm is established to study the six basic swinging and attacking actions of badminton under two types of grips. The test results show that the average recognition rate of the basic swinging movements under the two grips is 95.6 and 96.25%, respectively, and the overall recognition rate of the research model is 95.93%, which is 16.36% higher than that of the unimproved SVM algorithm, and the recognition rate of the research algorithm is the highest compared with many related algorithms. The experimental results show that the research algorithm is able to complete the recognition of badminton swing attack action, and the model algorithm has a higher recognition rate, which is of great value in the research of badminton attack action analysis.
Figures - available from: Signal Image and Video Processing
This content is subject to copyright. Terms and conditions apply.
A preview of this full-text is provided by Springer Nature.
Learn more
Content available from Signal Image and Video Processing
This content is subject to copyright. Terms and conditions apply.
Signal, Image and Video Processing (2025) 19:280
https://doi.org/10.1007/s11760-025-03865-7
ORIGINAL PAPER
SVM-Adaboost based badminton offensive movement parsing
technique
Chun-Yao Shih1·Yong-Tao Lin1·Wei Chen2·Jui-Chan Huang3
Received: 7 November 2024 / Revised: 27 December 2024 / Accepted: 23 January 2025 / Published online: 13 February 2025
© The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025
Abstract
With the development of smart devices and network technology, sports data statistics systems such as smart bracelets have
also been widely studied, but there are fewer studies on racket sports data analysis systems such as badminton and other
sports with complex movements, so this study will take badminton as the object of study, choose the inertial sensor-based data
acquisition method to collect real-time data, choose a combination of sliding window and action window to carry out data
segmentation, and use the principal component analysis is used to achieve data dimensionality reduction, and finally a two-
layer classification algorithm based on support vector machine and adaptive boosting algorithm is established to study the six
basic swinging and attacking actions of badminton under two types of grips. The test results show that the average recognition
rate of the basic swinging movements under the two grips is 95.6 and 96.25%, respectively, and the overall recognition rate
of the research model is 95.93%, which is 16.36% higher than that of the unimproved SVM algorithm, and the recognition
rate of the research algorithm is the highest compared with many related algorithms. The experimental results show that the
research algorithm is able to complete the recognition of badminton swing attack action, and the model algorithm has a higher
recognition rate, which is of great value in the research of badminton attack action analysis.
Keywords SVM-AdaBoost ·Badminton ·Action recognition ·PCA
1 Introduction
With the rapid development of big data on the Internet, sports
data in sports is more and more important to people, and
the application of smart devices for sports data collection
such as smart helmets and smart sports bracelets is more
and more widely used, but there are fewer studies related to
sports data analysis in racket sports such as badminton, so this
research will take badminton as the research object to carry
out the research on the badminton swing action recognition
and parsing system [1]. Aiming at the traditional image-based
BJui-Chan Huang
hjc0718@nkust.edu.tw
1Department of Physical Education, Hubei Polytechnic
University, Huangshi 435003, China
2Department of Physical Education, Shanghai University,
Shanghai 200438, China
3Department of Industrial Engineering and Management,
National Kaohsiung University of Science and Technology,
Kaohsiung City 807618, Taiwan
action recognition method which will be affected by the back-
ground colour of the picture, the occlusion and the light, etc.,
and is not applicable to the more complicated badminton
sports [2], and the research selects the inertial sensor-based
action recognition method for the study. Due to the data col-
lected by inertial sensors have the influence of non-action
and multi-dimensional data, the research selects window
interception method to segment the data, selects Principal
Component Analysis (PCA) to reduce the dimensionality of
the data, and selects Adaptive Boosting (AdaBoosting) and
Supported Boosting (AdaBoost) to reduce the dimensional-
ity of the data, and selects AdaBoost and Supported Boosting
(AdaBoost) to reduce the dimension of the data. AdaBoost
(AdaBoost) and Support Vector Machine (SVM) classifi-
cation and recognition methods are used to identify and
analyse the common badminton attacking movements. The
research aims to enable athletes to more accurately grasp their
technical characteristics, strengths, and weaknesses through
efficient analysis of offensive movements, assist coaches in
real-time evaluation of athletes’ offensive movements, timely
detection and correction of errors, and develop targeted plans
to improve training effectiveness and competition success
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Exploring Supervised Learning Methods for Predicting Cuisines from Their Ingredients
Article
Full-text available
  • Mar 2025
Various regions around the world use similar ingredients for food preparation, with exceptions of unique regional ingredients. However, the variation in the cuisines in the regions stems from the unique combinations of these ingredients. This aspect has been explored in Kaggle's competition, in which many submissions and solutions have been put forward. However, to the best of our knowledge, there is still no paper that compares Backpropagation, Support Vector Machine, Naïve Bayes, Decision Tree, Random Forest, and AdaBoost to predict cuisines based on their ingredients. We present our approach and measurement of those Supervised Learning Methods for tackling the problem. We use a combination of Machine Learning library and our own method implementations to conduct the experiment. Our results show that all the methods have more than 55% accuracy, and the best result achieved is 76.769% for Support Vector Machine. Given the small data size and high dimensionality of text data, SVM and Naive Bayes generalize well, compared to the more complex methods such as Neural Network. Our results also suggest that Random forest is robust and handles noise in the data well compared to AdaBoost.
View
Spam Detection Using Bidirectional Transformers and Machine Learning Classifier Algorithms
Article
Full-text available
  • Apr 2022
Spam email has accounted for a high percentage of email traffic and has created problems worldwide. The deep learning transformer model is an efficient tool in natural language processing. This study proposed an efficient spam detection approach using a pretrained bidirectional encoder representation from transformer (BERT) and machine learning algorithms to classify ham or spam emails. Email texts were fed into the BERT, and features obtained from the BERT outputs were usedto represent the texts. Four classifier algorithms in machine learning were employed to classify the features of the text into ham or spam categories. The proposed model was tested using two public datasets in the experiments. The results of the evaluation metrics demonstrate that the logistic regression algorithm achieved the best classification performance in both datasets. They also justified the efficient ability of the proposed model in detecting spam emails.
View
Research on Face Intelligent Perception Technology Integrating Deep Learning under Different Illumination Intensities
Article
Full-text available
  • Jan 2022
Aiming at the problem of face recognition under different illumination intensities combined with deep learning algorithm, this research designs a new loss function, i-center loss function and integrates the structure of migration learning algorithm on the basis of LeNets++ deep learning network. The face image data set labled faces in the wild with different illumination intensities and the image data set of supermarket monitoring system are used to train and test the improved LeNets++ deep learning network based on softmax, center and i-center loss function, and a variety of common image recognition networks. The calculation results show that although the amount of data required for the training of LeNets++ deep learning network is much larger than other networks selected in the study, when the loss function is changed to i-center, the accuracy of face image recognition under different light intensities is significantly improved, reaching 99.65%. In the supermarket data set, the maximum face recognition rate of the algorithm using i-center loss function is 99.07%, which is 0.21% and 0.6% higher than that of using center softmax and softmax loss function, respectively. Therefore, experiments show that the improved deep learning neural network based on i-center loss function can improve the effect of face recognition under different illumination intensities.
View
Research on Microvideo Character Perception and Recognition Based on Target Detection Technology
Article
Full-text available
  • Jan 2022
With the rapid development of the Internet short-video platform, microvideo is increasingly attracting users because of its small traffic and time period. However, the content of microvideo on the Internet is complex, with illegal information. This research is an improved algorithm based on target detection technology, aiming at the characteristics of difficult character recognition in microvideo. The algorithm combines Gabor wavelet transform algorithm and 2D principal component analysis (PCA) algorithm and makes full use of video perception technology to realize character recognition in microvideo. Through MATLAB simulation analysis, it can be seen that the accuracy of the optimization algorithm proposed in this paper is 86.34% and the recognition time is 34.28s. Compared with the traditional PCA algorithm and artificial neural network (ANN) algorithm, the optimization algorithm proposed in this paper has better recognition rate and recognition efficiency.
View
Image Recognition of Badminton Swing Motion Based on Single Inertial Sensor
Article
Full-text available
  • Sep 2021
This article analyzes the method of reading data from inertial sensors. We introduce how to create a 3D scene and a 3D human body model and use inertial sensors to drive the 3D human body model. We capture the movement of the lower limbs of the human body when a small number of inertial sensor nodes are used. This paper introduces the idea of residual error into the deep LSTM network to solve the problem of gradient disappearance and gradient explosion. The main problem to be solved by wearable inertial sensor continuous human motion recognition is the modeling of time series. This paper chooses the LSTM network which can handle time series as well as the main frame. In order to reduce the gradient disappearance and gradient explosion problems in the deep LSTM network, the structure of the deep LSTM network is adjusted based on the residual learning idea. In this paper, a data acquisition method using a single inertial sensor fixed on the bottom of a badminton racket is proposed, and a window segmentation method based on the combination of sliding window and action window in real-time motion data stream is proposed. We performed feature extraction on the intercepted motion data and performed dimensionality reduction. An improved Deep Residual LSTM model is designed to identify six common swing movements. The first-level recognition algorithm uses the C4.5 decision tree algorithm to recognize the athlete’s gripping style, and the second-level recognition algorithm uses the random forest algorithm to recognize the swing movement. Simulation experiments confirmed that the proposed improved Deep Residual LSTM algorithm has an accuracy of over 90.0% for the recognition of six common swing movements.
View
View
View
Separation of overlapping fingerprints by principal component analysis and multivariate curve resolution—alternating least squares analysis of hyperspectral imaging data
Article
  • Jan 2022
Overlapping fingerprints are often found at crime scenes, but only individual fingerprints separated from each other are admissible as evidence in court. Fingerprint components differ slightly among individuals, and thus their fluorescence spectra also differ from each other. Therefore, the separation of overlapping fingerprints using the difference of the fluorescence spectrum was performed with a hyperspectral imager. Hyperspectral data (HSD) of overlapping fingerprints were recorded under UV LED excitation. Principal component analysis (PCA) and multivariate curve resolution—alternating least squares (MCR–ALS) were applied to the HSD to determine the optimal method for obtaining high-contrast images of individual fingerprints. The results suggested that MCR–ALS combined with PCA-based initialization is capable of separating overlapping fingerprints into individual fingerprints. In this study, a method for separating overlapping fingerprints without initial parameters was proposed.
View
Motion recognition technology of badminton players in sports video images
Article
  • Jun 2021
  • FUTURE GENER COMP SY
Intelligent identification technology has made great progress in transportation, and its application in sports has attracted widespread attention. This research mainly discusses the research of human motion recognition technology in sports dance video images. In geometric algebraic space, using instance templates, a new cfrdF method, using css-based similarity of human body features as features, to construct an angle-adaptive and continuous-scale space template matching algorithm to calculate the similarity between horizontal plates and detected images value, set a certain threshold, so as to match the area where the human body is located. Based on the actual video analysis and display, using the self-similar structure of the color of the human object as the basic feature, it is described by the geometric algebra method, and the human body in the video image is extracted using template convolution objects and design adaptive template functions to extract human movements from sports dance video images. By loading each test video in turn, the posture of the characters in the video is checked every 30 s, and the approximate positions of the head and feet are marked and output to a txt file. Because lying on the ground during sports dance is a way of matching by height correction, the accuracy is less disturbed, but it can still reach 90.9%. The research results show that the template matching robot model proposed in this paper can accurately and robustly extract human objects in video.
View
Implementation of support-vector machine algorithm to develop a model for electronic commerce energy regulatory system
Article
  • Nov 2021
In the current stage, China has paid special focuses to the data and information utilization of the Electronic-commerce (e-commerce) energy regulatory system. However, it is difficult to guarantee the scientific data information management of the energy regulatory system at present. It was considering how to use the scientific framework for data mining has been the recent hot spot. Under this condition, this manuscript analysis the e-commerce energy regulatory system model based on data mining. This paper begins with the discussions on the current models of the energy regulatory system at home and abroad, enhances the methodology combined with the shortage of data mining, and then applies support-vector machine (SVM) algorithm in the e-commerce energy regulatory model. The experimental analysis shows that the revised SVM algorithm can realize the objective evaluation of regulatory efficiency based on data mining and can lead to the optimal strategy according to the scenarios in the actual application steps of the energy supervision system. The performance is satisfactory, which means the energy supervision system can reach more than 97%, this performance is better than most of the latest approaches.
View
Establishment of naive Bayes classifier-based risk prediction model for chemotherapyinduced nausea and vomiting
Article
  • Apr 2021
Objective: To establish a risk prediction model of chemotherapy-induced nausea and vomiting based on naive Bayes classifier. Objective: We collected the basic information, treatment protocols and follow-up data from 300 patients receiving chemotherapy in the Oncology Department of Second Xiangya Hospital from July to September, 2020. Correlation analysis was carried out between the potential factors related to nausea and vomiting in the treatment plan and the individual characteristics of the patients. For the two characteristics with a correlation coefficient greater than 0.8, their contribution to the area under curve (AUC) was calculated, and the characteristic with a smaller contribution was removed. The naive Bayes classifier in the machine learning library scikit-learn was used as the prediction model of chemotherapy-induced nausea and vomiting, and 10-fold stratified-shuffled-split cross-validation was used to obtain the final result of the model. The machine learning model was trained using 70% of the samples, and 30% of the samples were used as the test set to assess the performance of the model. Objective: The sensitivity of the model for predicting the risk of nausea and vomiting due to acute chemotherapy was 0.83±0.04 (95%CI: 0.80-0.86) with a specificity of 0.45±0.03 (95%CI: 0.42-0.47) and an AUC of 0.72±0.04 (95% CI: 0.69-0.75). The sensitivity of the model for predicting the risk of delayed chemotherapy-induced nausea and vomiting was 0.84±0.01 (95%CI: 0.83-0.86) with a specificity of 0.48±0.03 (95%CI: 0.45-0.52) and an AUC of 0.74±0.02 (95%CI: 0.72-0.77). Objective: The naive Bayes classifier model has a good performance in predicting the risk of chemotherapy-induced nausea and vomiting in Chinese cancer patients.
View