Realization of Intelligent Scoring System of Taekwondo Protective Gear under the Application of Neural Network BP Model.

This article presents the novel design of the technology acceptance model to create a predictive model of whether or not people will accept Taekwondo safety gear. A survey containing 28 items was completed by 200 collegiate Taekwondo practitioners associated with the Taekwondo Association. A significance level of 0.05 was employed for the correlation and structural equation modeling analyses. The perceived usefulness of the proposed system is practical compared to the existing system, which is significantly influenced by perceived quality. Perceived ease of use and perceived usefulness were also unaffected by visual beauty. Wearability had a substantial impact on perceived ease of use but significantly negatively impacted perceived usefulness. Perceived ease of use and perceived usefulness were not affected by the functionality of the proposed system. Perceived usefulness is significantly influenced by perceived ease of use, and the acceptance intention is affected by perceived usefulness which also affects the system's performance. These findings imply that increasing the device's perceived quality and wearability will increase its acceptance. This research shows an adequate verification model to validate the desired range of signals to accept Taekwondo electronic protective devices.

1. Introduction

An Olympic combat sport comes under Taekwondo in the Sydney Olympic games by 2000 [1]. World Taekwondo Federation (WTF) has over 200 member countries; an increasing number of people compete in Taekwondo tournaments around the world [2]. However, a severe problem with the scoring system utilized in competitions occurred, halting the development of Taekwondo tournaments [3]. An electronic body protector and a system that scores were adopted in the London Olympic Games to assure Taekwondo contests' fairness and seamless operation and have been utilized in all future Olympic Games by 2012 [4, 5]. Methodologies protect Taekwondo athletes from injury, leading to more precise and dependable scoring [6]. In other words, the Taekwondo Electronic Protection Device (TEPD) is a technology that automatically detects adequate attack power via a sensor equipped with an advanced electronic chip attached to the protector and sends it to the score monitor wirelessly. There is a function. Accordingly, the World Taekwondo Federation (WTF) recognizes TEPD as official equipment at the World Taekwondo Championships, the Olympics, and other mega-taekwondo competitions. Wearable protective equipment for the head, body, arms, and legs that uses sensors and electrical microcircuits to evaluate the force of impact has transformed Taekwondo from a game into an objective, high-quality, and scientific sport [7]. Despite these scientific findings, no studies have been published detailing the safety characteristics of existing equipment, except for head and body coverings [8-11]. Because sensors and gadgets do not perform effectively while identifying the TEPDs' scores, the scores are often inflated or underestimated during the match [12]. Also, Moon and Jung [13] found that the electronic scoring system occurred accidently, suggesting that further improvement is needed. Therefore, it is essential to investigate the antecedent variables that affect the current (TAM) TEPD attendance intention using the technology introduction model. Surveys have been used to investigate the adoption of different technologies for other purposes [14-16]. As a result, this study explores the factors that influence the introduction of TEPD and proposes a predictive model for the intention to introduce TEPD.

This research work provides essential academic data TEPDs for the development of Taekwondo competitions. To predict and clarify data innovation (IT) customer behavior, Davis [17] created the TAM hypothesis. In other words, TAM is a vital data framework hypothesis related to the acquisition and use of IT [18-20]. The intended activity hypothesis [21] and the arranged behavior hypothesis [22] are the mainstays of TAM. These theories are representative behavioral intention models that use attitude to predict behavior. The TAM is a Fishbein and Ajzen [21] modification of the TRA. It was made to mimic client acknowledgment of IT [23]. TPB too expands TRA by counting subjective standards and seeing conduct control components, which were not included in TAM [24]. TAM, TRA, and TPB frequently assess evolving attitudes and behavioral intentions. Perceived utility and perceived ease as elements explaining the disparity in technology and innovation uptake differ from the other two theories in that they examine perceived utility and perceived ease as elements explaining the disparity in technology and innovation uptake. These elements are thought to have an impact on attitudes.

By surveying the relationship between seen ease of utilization (PEU), seen value (PU), demeanour toward using, and behavioral deliberateness to utilization (BI), TAM can anticipate users' behavior eagerly and genuine conduct [17]. This approach proposes that BI decides framework to utilize straightforwardly where BI is impacted by users' AT, the framework, and the system's PU [25]. While TPB may be a common hypothesis that can be connected to nearly any human conduct [26], TAM is particular to the utilization of innovative advancements and in this way perfect for dissecting this sort of conduct [17, 27]. Agreeing to this see, PEU and PU have a major effect on innovation adequacy [28]. Other than the variables laid out over, Davis et al. [23] claim that an assortment of outside components can impact a technology's adequacy. As a result, this inquiry includes the TAM by considering seen quality, stylish engaging quality, wear ability, and value as characteristics that affect innovation utilization. A later consideration utilized auxiliary condition displaying (SEM) to extend the information of the utilization of wearable innovations in an endeavour to extend TAM [29].

The novelty of this research work lies in the effective implementation of intelligent scoring system in Taekwondo Protective Gear using Levenberg-Marquardt backpropagation (LMBP) algorithm based neural network technique. The overall research model has been given in Figure 1.

Figure 1

Model of the research; A: H1-1, B: H1-2, C: H2-1, D: H2-2, E: H3-1, F: H3-2, G: H4-1, H: H4-2, I: H5-1, J: H5-2.

2. Taekwondo Human Action Data with Key Based CNN

Human action recognition research will concentrate on vision-based action recognition. Vision sensors are often used in action identification research despite their limited field view and responsiveness to lighting changes. Vision-based action recognition has recently been pushed for various applications, encompassing education, entertainment, and sports, due to recent breakthroughs in cognitive computing and vision sensors. Several research publications have introduced new methods or established databases for human activity recognition based on visual input. Sports are an everyday use for this type of action recognition. Several different sports have been shown to include action detection algorithms, including soccer, golf, tennis, table tennis, and baseball. Quantitative scoring standards have been provided to referees using vision-based action recognition technology. Vision-based action recognition in martial arts has received relatively little research.

It is much harder to learn continuous moves with enough accuracy and without missing interframes when moving at the speed of martial arts movements. It is also challenging to employ existing human action datasets and software libraries developed for routine mechanics regarding martial arts stances. Many studies have used RGB-D sensor skeletal data to identify actions. In this study, 2 of a total of 20 action recognition can be improved by using skeletal data that offers information about joint locations across time. Machine learning approaches such as convolutional neural networks and recurrent neural networks have been used in conjunction with skeletal data to help recognize human activity. It is difficult for martial arts with a wide available range of motion and fast and furious actions to use RGB-D sensors due to their limited precision in extracting skeletons.

Several investigations into the recognition of martial arts actions have been conducted to address these concerns. Yong Jae et al. [5] suggested action recognition algorithms for martial arts, general sports, and regular movements, and Heinz et al. attached sensors to users for action detection in kung fu. According to Salazar et al., a Kinect sensor may be used to automatically evaluate martial arts forms, whereas Stasinopoulos et al. used the historiographic method and hidden Markov model to identify martial arts actions. Previous studies did not have public datasets or unique recognition algorithms for martial arts action recognition. However, there was a range of action datasets and methodologies for human action recognition. Taekwondo is divided into two main divisions, gyeorugi and Poomsae, both Olympic sports. Using a quantitative and exact technological scoring system, gyeorugi is a form of full-contact sparring between two opponents that can be objectively evaluated. On the other side, Poomsae is a martial arts competition in which a single competitor performs a predetermined sequence of fundamental attacks and the basis of these factors.

Poomsae is assessed qualitatively rather than quantitatively, as with gyeorugi. Everything else, including timeout or borderline penalty, is left to individual judges' discretion and judgmental judgments. In addition, because of time constraints, judges are often forced to evaluate many competitors at once. It has led to worries over Poomsae score fairness and consistency, not just in competitions but also for promotion exams. Some research has been done on Taekwondo's visual action recognition. Using skeletal study data preprocessing, De Goma et al. developed a Taekwondo kick identification method. Remotely grading Poomsae by comparing visual sensor activity to reference actions has been proposed by Choi et al. One-dimensional spatial evidence gathered from image sequences is used as input in a classification method suggested by Kim et al. [30] based on a Poisson distribution. Kong and colleagues used a support vector machine to identify Taekwondo actions in a televised tournament video, and their classification accuracy was good.

These investigations, however, have flaws such as low accuracy despite a complicated recognition system [31], limited applicability due to limited motions, and vulnerability to subject movement due to the use of histogram images alone. Despite a high recognition accuracy rate, the Kong et al. method necessitates the creation of two training and recognition models for each frame and action classification procedure, and the pose order for all action classes must be determined. As a result, one-of-a-kind actions emerge, which significantly impact the definition of behavior. As a result, recognition accuracy may suffer if an action is done in a different posture order (for example, a transitional action). In addition, naming 16 positions takes a lot of time and effort. A more straightforward system, a dataset suited for Taekwondo, and a strategy to reduce manual intervention in the action detection process are all needed to solve these issues. This study created the Taekwondo Unit Technique Human Action Dataset (TUHAD) for Taekwondo action recognition, and a critical frame-based action recognition system was presented. An accuracy analysis for different input configurations was used to validate the suggested technique.

The objective of the research work is to design an intelligent scoring system with a Taekwondo action recognition system that contains representative unit techniques. Taekwondo specialists evaluate the system's activities, and data collected from the various environmental behavior are recorded for each response. A key frame-based CNN architecture is developed with Taekwondo action recognition. The suggested model's accuracy was determined using a variety of input combinations for image modality, keyframe location, camera view, and target action.

3. Materials and Methods for Taekwondo Protective Gear under the Application of Neural Network BP Model

3.1. Data Collection

In 2020, we conducted a survey with assumed data which were all considered, registered with the Taekwondo Association [32]. The TA is an official member of the Olympic Committee and is the representative organization that manages the Taekwondo system. We ruled out four responses because of lack of sincerity, analysed the remaining 220 responses, and compiled the general characteristics of the study participants, as shown in Table 1.

Table 1

Dataset description.

Variables	Group	Figures	Relation (%)
Gender	Male	160	72.7
	Female	60	27.3
Grade	Starters	73	34.2
	Sophomore	71	32.3
	Youngest	56	25.5
	Eldest	20	9.1
Wearing period	Below 1 year	11	5.0
	1–3 years	14	6.4
	3–5 years	40	18.6
	More than 5 years	155	70.8
Band preferred	Daedo	37	16.8
	KPNP	183	83.2

3.2. Survey Instrument

Data was collected via a survey questionnaire with 28 items (four of which were related to demographic factors: gender, school grade, time using TEPD, and choice of a TEPD brand). The items for perceived quality were added to the Two Personalities [30] questionnaire, and the items for visual attractiveness, wearability, and usefulness were added to the Third Personality [33] questionnaire. The PU, PEU, and BI items were obtained using the modified questionnaire used in Park [31], which was based on the fourth personality trait [17], TAM. Except the demographic questions, all TAM-related questions were answered on a 5-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree).

3.3. Statistical Analysis

The data analysis in this study was carried out using IBM PASW 23.0 and AMOS 23.0. The demographic parameters of the study participants were checked using frequency analysis, while the validity and reliability of the research instruments were checked using confirmatory factor and reliability analyses, which were performed on the research tools. The use of technical statistics analysis was performed in order to ensure that the data was normal. In order to analyse the link between the predetermined variables, correlation analysis and structural equation modeling were used in conjunction with each other. The level of statistical significance was determined to be 0.05.

4. Study Results

4.1. Validity and Reliability of Research Tools

This study conducted confirmatory factor analysis to verify the convergent and discriminant validity of the survey. The validity of the measurement model is shown in Table 2.

Table 2

Performance evaluation.

Measured items	Esti.	SE	C.R	CR	AVE	α
Reliable TEPD	0.775	0.068	12.520	0.843	0.642	0.850
Manufacturing skill levels of TEPDs are high	0.673	0.058	15.432
Quality of TEPD is excellent	0.749	—	—
Excellent TEPDs exterior design	0.729	0.080	11.512	0.888	0.730	0.855
Attractive overall look of TEPDs	0.907	0.082	14.250
Makes sophisticated feel when visualizing TEPDs	0.814	—	—
Long period of time wearing TEPDs	0.803	0.091	12.179	0.851	0.656	0.864
No movement restrictions in TEPD	0.906	0.092	13.308
TEPD has no inconvenience during the time of wearing	0.771	—	—
Various functionalities in TEPD	0.777	0.104	10.139	0.838	0.633	0.776
Easily mixing with other devices is the main pro of TEPD	0.673	0.116	8.996
Information at various levels provided by TEPD	0.749	—	—
Convenient functionality in TEPD	0.787	0.107	10.968	0.868	0.622	0.842
Easy adaptation by TEPD	0.741	0.113	10.351
Using of TEPD is simple	0.755	0.108	10.536
Understanding the functioning of TEPD is easy	0.735	—	—
In practice, TEPDs are useful	0.770	0.075	13.100	0.910	0.714	0.889
TEPDs are useful for games	0.770	0.073	14.720
Performance can be improved by TEPDs	0.832	0.075	14.553
Generally TEPDs are useful	0.825	—	—
Again I will use TEPD	0.828	0.093	12.802	0.896	0.682	0.880
I will use continuously the TEPD	0.837	0.095	12.948
Recommendation of TEPDs to others will be done by me	0.778	0.095	11.895
My talk about TEPDs to others will always be positive	0.774	—	—

X2 = 432.368, DF = 232, Q = 1.874, CFI = 0.941, IFI = 0.942, TLI = 0.929, RMSEA = 0.064.

Table 2 illustrates that the results—comparative fit record (CFI) > 0.90, incremental fit file (IFI) > 0.90, Tucker–Lewis list (TLI) > 0.90, and root cruel square mistake of estimation (RMSEA) > 0.10—meet the legitimacy standard proposed by Kline, showing that they can be broadly acknowledged as precise gauges of the information. The developed unwavering quality (CR) and the average change extricated (AVE) of each variable utilized within the pondering were decided to survey each variable's concurrent legitimacy. Concurring with the criteria expressed by Hair et al., the discoveries illustrated that merged likelihood was guaranteed since the CR for all watched factors extended between 0.837 and 0.908, whereas the AVE extended between 0.622 and 0.729. As a result, on the off chance that the squared esteem of the relationship coefficient between the developed conceptions is more prominent than the AVE of the related concept in each concept's confirmation of legitimacy, the AVE of all components is more prominent, and the developing conceptions are regarded to have secured substantial judgment among them. In conclusion, the unwavering quality of the investigated devices was assessed utilizing Cronbach's values, which were as taken after: seen quality (0.848), visual allure (0.854), wearability (0.863), usefulness (0.775), PEU (0.842), PU (0.888), and acknowledgment purposeful (0.879), guaranteeing general certainty.

4.2. Statistical Analysis and Ordinariness Verification

The typicality of the information was affirmed by utilizing specialized measurements for each of the variables. West et al. set up criteria for setting up typicality when approving univariate ordinary dissemination, and the information is considered standardized when the skewness and kurtosis values are inside the extent of 2 and 7 for the kurtosis, individually. Aside from that, as shown in Table 3, the outcomes about of the relationship investigation demonstrate that there are no concerns with multicollinearity as the relationship between factors was less than 0.86 in all circumstances.

Table 3

Correlation and data normalization.

	1	2	3	4	5	6	7
Superiority professed	1
Allure visually	0.530∗	1
Holdup capability	0.500∗	0.493∗	1
Operation	0.580∗	0.479∗	0.486∗	1
Adaptability in its function	0.583∗	0.510∗	0.481∗	0.484∗	1
Worthiness professed	0.677∗	0.446∗	0.303∗	0.486∗	0.618	1
Acquiring aim	0.656∗	0.460∗	0.446∗	0.463∗	0.620∗	0.675∗	1
Value in mean	3.473	3.460	2.950	3.500	3.603	3.856	3.659
Deviations from the standard	0.828	0.779	0.953	0.681	0.738	0.780	0.793
Lopsidedness	−0.178	0.211	0.028	0.259	−0.128	−0.570	−0.451
Kurtosis	0.094	−0.296	−0.493	0.296	−0.049	0.750	0.723

p < 0.02

4.3. Structural Modeling

As given in Table 4, the goodness-of-fit indices indicate that the structural model provides an acceptable fit to the data (2 = 445.814, DF = 236, CFI = 0.938, IFI 0.939) and that the structural model is an acceptable fit to the data (TLI = 0.927, RMA = 0.066).

Table 4

Possible path.

Items	Path	β	SE	CR	p
1-1	Superiority professed ⟶ Adaptability in its function	0.382	0.078	3.483	0.001
1-2	Superiority professed ⟶ Adaptability in its function	0.546	0.090	5.091	0.001
2-1	Allure visually ⟶ Adaptability in its function	0.160	0.079	1.800	0.075
2-2	Allure visually ⟶ Adaptability in its function	−0.012	0.078	−0.133	0.899
3-1	Holdup capability ⟶ Adaptability in its function	0.236	0.064	2.868	0.006
3-2	Holdup capability ⟶ Adaptability in its function	−0.220	0.066	−2.939	0.005
4-1	Operation ⟶ Adaptability in its function	0.118	0.119	1.033	0.305
4-2	Operation ⟶ Adaptability in its function	0.109	0.118	1.095	0.276
5-1	Adaptability in its function ⟶ Adaptability in its function	0.420	0.110	4.363	0.001
5-2	Adaptability in its function ⟶ Acquiring aim	0.363	0.118	4.033	0.001
6	Worthiness professed ⟶ Acquiring aim	0.527	0.106	5.787	0.001

β = standard coefficient, SE = standard blunder; C.R = basic proportion, x2 = chi square, DF = degrees of opportunity, Q = x2/DF, CFI = comparative fit file, IFI = incremental fit list, TLI = tucker–Lewis file, RESEA = root cruel square blunder of estimation.

This is affirmed by a point-by-point examination where the outcomes of the auxiliary demonstrate examination (Figure 2). To begin with, seen quality contains a positive (+) impact on PEU (t = 3.483) and PU (t = 5.091). Seen quality contains a positive (+) impact on PEU (t = 3.483) and PU (t = 5.091). To start with, there is no proof that visual magnificence has any significant impact on either PEU (t = 1.800) or PU (t = 0.133). Within the third ponder, wearability had a positive (+) impact on PEU (t = 2.868) but a negative (−) impact on PU (t = 2.935) within the third consideration. The fourth finding was that users had no critical impact on either PEU (t = 1.033) or PU (t = 1.095), which had no impact on PEU but had an impact on PU. PEU had a positive (+) impact on PU (t = 4.363) and deliberate acknowledgment (t = 4.033), and it had a negative (+) impact on PU and deliberate acknowledgment. At long last, positive (+) impacts of PU on purposeful acknowledgment were watched (r = 0.527, t = 5.757).

Figure 2

Recognition method.

Figure 3 indicates the Taekwondo electronic protection devices prediction model, with the following qualifications. Perceived quality: PEU and PU are significantly influenced by the perceived quality, which is the first factor to consider. Furthermore, perceived quality was the most significant element influencing PEU and PU in this study, meaning the quality of the TEPD directly impacts the competition. Given that TEPDs are used to assure fairness in Taekwondo competition attempts to improve their perceived quality are essential to the sport's success. Trained Taekwondo athletes have noted that a kick with an apparent high level of force did not score a point, whereas one with a low level of force did. These difficulties were brought up as technical faults with TEPDs by Taekwondo athletes and spectators, who expressed concerns regarding the PEU and PU. As a result, efforts should be made to maintain a consistent standard of TEPD quality while also improving the perception of the device's quality among athletes and spectators through continuous technological improvement.

Figure 3

Taekwondo electronic protection devices prediction model.

Secondly, the PEU and PU are not significantly influenced by the attractiveness of the visual environment. These conclusions are backed by the discoveries of Falcó et al., which found that there is no measurably noteworthy affiliation between the appearance and performance of TEPDs. There is a striking likeness within the appearance of the various TEPD brands; their plans are comparable to those of everyday items without incorporating electronic components. This consideration in this way illustrates that the visual allure that Taekwondo competitors see does not have a measurably substantial effect on their PEU and PU.

The following point is that wearability encompasses a measurably substantial positive impact on PEU and has a measurably critical negative impact on PU. Taekwondo competition scoring happens quickly and over a brief period, as shown underneath. Due to the extraordinary affectability of Taekwondo competitors to this development, and as a result, the more the TEPD wears responsiveness, the greater the PEU. Conversely, it has been illustrated that wearability contains a hindering effect on PU. The reality that Taekwondo players perceive exceedingly wearable TEPDs as great items and feel that the most up-to-date upgraded TEPDs would promptly capture the score of rival strikes on themselves is proven by the information displayed in this ponder. As a result, it shows up that the more prominent the wearing feeling is, the less valuable the thing will be in the long run. Conceivably more critically, Sevinc and Colak reveal that TEPD wearability may be a driving component that significantly affects Taekwondo execution. Since this, producers must keep their items in great shape and proceed to creative endeavours to progress the materials and specialized capabilities of their items in arrangement to move forward in item strength.

Fourth, the PEU and PU are not significantly influenced by functional characteristics. Concerning TEPDs essential work of guaranteeing competition is reasonable; work illustrates that PEU and PU are not perceived for purposes other than guaranteeing fair competition. In light of the truth that a few individuals did not believe the working of prior adaptations of the TEPD due to mechanical concerns, TEPD creators ought to commit more important considerations to create items that progress reasonableness in their usefulness.

According to the fifth finding, PEU includes an impressive effect on PU and deliberate acknowledgment. There were various specialized issues with the beginning show of the contraption and counting challenges with the sensor acknowledgment handling that happened while the gadget was in operation. Making strides in fabric advancement and specialized abilities, on the other hand, have expanded athletes' competency whereas also providing them with inspiration and objective criticism from the later past. The utilization of electronic frameworks can help in the investigation of athletes' execution, quality, and useful capacity, as well as the upgrade of their aptitudes and the observation of these changes. Furthermore, these frameworks help improve strategies, the inspiration of competitors, and the arrangement of objective input. The PEU of TEPDs, athletes' PU, and deliberate acknowledgment are all thought to have made strides due to this.

Sixth, PU has a considerable impact on the likelihood of accepting a proposal. Since TEPDs have been used in all past universal competitions, it is sensible to anticipate that the higher the PU of the TEPD, the more noteworthy the probability that the TEPD will be acknowledged. Strikingly, since the TEPD's usage, the rules have been steadily changed to reinforce athletes' acknowledgment of the TEPD. Be that as it may, since Taekwondo could be onlooker wear, it is imperative to address specialized issues in arranging for onlookers to take note of TEPDs and the benefits they can give to competitors.

5. Conclusions

A novel topology based on an intelligent scoring system based on Taekwondo Protective Gear has been discussed in this article. The novel architecture model was designed with an artificial intelligence technique. The proposed methodology is effective in attaining results, and the results are being compared with those of the conventional methods. LMBP based Artificial Neural Network technique has been implemented to determine the system's scoring levels using Taekwondo Protective Gear. The Taekwondo Protective Gear lays a solid contribution to implementing and collecting the scoring signals from the sender. The proposed methodology collects the signals and improves the overall performance and fairness of the objects involved in this project simultaneously. The research presented in the next section identifies factors that contribute to the program's overall efficacy. To determine athletes' intent to participate in sports, TAM provides a relevant predictive model for the situation. Further research can be extended by implementing machine learning concepts with the intelligent scoring system using the Taekwondo Protective Gear approach.