Impact of Intelligent Management of Physical Education Teaching and Athletic Training Integration on Improving Teaching Quality

Sport is produced in the process of human production and life. Like all other human cultures, its various characteristics are reflected in the way humans behavior, so different sports systems have different performances in practical activities according to their different functional characteristics [1]. The main form of school sports is physical education teaching, while the main form of competitive sports is sports training. The different characteristics of physical education teaching and sports training in the process of practical activities reflect the different functional characteristics of school sports and competitive sports, understanding the respective characteristics of physical education teaching and sports training and their interrelationships, but also at the same time mastered the characteristics and interrelationships of school sports and sports training [2–3].

Throughout the history of the development of Chinese sports and the current situation of modern sports, it is not difficult to find that most of the research focuses on physical education and sports training, which is because they are closely connected with practice and are the main way to realize the many guidelines and policies in the development process of school physical education and competitive sports [4–6]. Although there are many different types of activities in school sports and athletics, physical education and athletic training are the most systematic and the best means of realizing the functions of school sports and athletics [7]. Unlike other sports activities in school sports and competitive sports, physical education and sports training have a distinct dominant nature, and human management and control make their activities relatively centralized, and the methods and contents of imparting knowledge and skills are relatively stable, so they have unique advantages in theoretical development and organizational management [8–9]. This makes physical education and sports training hold an irreplaceable position in realizing the functional value of school sports and competitive sports as the best expression of their functional characteristics [10–11]. Their advantages in realizing the functions of school sports and competitive sports determine that they can reflect the current situation of school sports and competitive sports in the most direct and real way, which makes them the focus of all research in school sports and competitive sports [12–13]. So it is said that physical education and sports training are the fundamental issues in the development research of modern school sports and competitive sports, and the research and discussion of intelligent management related issues of their combination will play a certain role in promoting the quality of school sports teaching [14–15]. 1)

Physical education

As an important part of school education, physical education plays an irreplaceable role in improving student’s physical fitness and promoting their physical and mental health. Literature [16] provides a comprehensive overview of physical education students’ and teachers’ attitudes toward physical education teaching from three aspects: reviewing the popular theoretical models in attitude research, describing the methodology used to locate the studies used in the rest of the paper, and discussing the measurement problems in attitude research, which provides a reference value for adjusting or innovating physical education teaching strategies. Literature [17] explores the meaning construction pedagogy in physical education through real-life case studies and finds the exploration of students’ meaning in physical education fascinating and informative, and the study has implications for the development of physical education for lifelong movement. Literature [18] investigated the level of motivational values and attitudes towards physical education as well as the level and dynamics of their physical fitness using students of Ukrainian medical, educational institutions as the object of the study, and the results of the investigation showed that the level of motivation of students for physical activity in physical education directly or indirectly affects the level of development of their physical fitness and the level of their physical abilities in their future professional activities.

2)

Athletic training

Currently, sports training in Chinese colleges and universities is dedicated to enhancing students’ physical fitness and pursuing excellent sports performance. Literature [19] examined the mechanisms of the effects of sports practice and functional training on body composition, body mass, cardiometabolic risk and psychological responses in female adolescents participating in a multidisciplinary program and found that both types of sports training were effective in improving a subset of obesity-related health parameters through the design of a relevant intervention experiment. Literature [20] synthesized human-computer interaction technology under information technology and artificial intelligence to design a set of intelligent sports management systems based on deep learning technology, which can provide students with a good sports training environment and meet the teaching needs of colleges and universities and has a certain effect in improving the quality of college students’ physical education and promoting psychoeducation. Literature [21] introduced VR technology in physical education and sports training, aiming to simulate sports scenes by generating VR panoramas to break through the limitations of traditional physical education courses in terms of sports venues, equipment, and safety and then effectively improve learners’ motivation and sports skills.

3)

Integration of physical education and sports training

Physical education and sports training have their focuses, but there is also a relationship of mutual complementation and co-promotion, and only by reasonably combining the characteristics and objectives of the two can students’ physical literacy be better developed. Literature [22] focuses on the understanding of the effects of physical education and athletic training and believes that the combination of physical education and athletic training will turn into one of the most powerful educational mediums, which can enhance the quality of teaching and learning, and is important for the promotion of students’ healthy physical and mental growth. Literature [23] takes Indonesian secondary schools as an example. It proposes a holistic educational framework combining a physical education model and athletic challenge approach based on the ecosystem theory, aiming to improve athletes’ physical fitness and to develop competent, educated and enthusiastic athletes. The feasibility of this integrated teaching model has been verified by experiments, which can help to enrich and improve the level of physical education in Indonesia. Literature [24] constructed an interactive teaching system for sports with the support of artificial intelligence technology, which mainly combines physical education and sports training in order to realize intelligent teaching management. The superior performance of the constructed system was verified through system testing, which enhances the integrity of the teaching process and improves the quality of physical education.

In this paper, the particle filter algorithm is used to track the human body’s movement postures in the process of sports training for physical education, and the key joint points of the human skeleton are used as the reference coordinates to achieve accurate tracking of the human body’s movement postures. The parameters in the pose estimation are optimally adjusted by constructing a particle swarm equation with a self-organising mechanism, and the sparse representation model is used to analyse the fine and coarse poses of the human body during sports training. Then, refer to the key action frames in the canonical sports action database from the key action frames in sports training and identify and process the irregular body movements in them. Finally, an intelligent management system combining physical education and sports training is designed on this basis, through the visual way to intuitively display and assess the various types of movements of students in sports training, to assist teachers to optimise the teaching programme, and to promote the improvement of the quality of physical education teaching. After testing the effectiveness of the human posture estimation method, this study applies the intelligent management system to a sports school and analyses whether the application of the system has a positive impact on the quality of physical education teaching in terms of students’ motor skills, sports cognition and teachers’ teaching ability.

2

Method

2.1

Human Posture Estimation Methods in Sports Training

2.1.1

Human body tracking algorithm

This paper proposes the particle filtering algorithm (PF) [25] to track the human body posture of students in sports training in physical education, which can deal with the non-Gaussian characteristics in the probability distribution in the tracking environment. On this basis, the coordinate value (x, y) of the neck of the human skeleton is chosen as a representative reference value of the human body in sports training, because the neck joint is the key point that is most easily located by the human posture estimation algorithm, and a pentagonal region is formed by taking the neck as the vertex and the joint points of the shoulders, the joint points of the left and right hipbones, and the tracking of the personnel is carried out according to the RGB information of this pentagonal region. The specific principle is as follows, assuming that a state variable X is defined: 1 $X = {[\begin{matrix} x, \dot{x}, y, \dot{y}, r, g, b \end{matrix}]}^{T}$

The prediction and observation models for the particle filter model are: 2 $x [\begin{matrix} t \end{matrix}] = A x [t - 1] + v, v ~ ζ (O, {0, σ_{x}, 0, σ_{y}, σ_{c}, σ_{c}, σ_{c}}) \in R^{7}$ 3 $y [t] = C x [t]$ 4 $A = d i a g (A_{s}, A_{s}, I_{3}), A_{s} = [\begin{matrix} 1 & Δ \\ 0 & 1 \end{matrix}]$ 5 $C = d i a g ([\begin{matrix} 1 & 0 \end{matrix}], [\begin{matrix} 1 & 0 \end{matrix}], I_{3})$

In Equation (3) t is the time counter, A is the state transfer matrix, v is the position random shift variable, and σ_* is the standard deviation parameter, also known as the tuning parameter in particle filter algorithms. In Equation (4), y is the observation variable. In Equation (5), △ is a fixed time constant of about 1/29. 97 seconds. The value of σ_* has to be changed as the viewing angle and distance of observation change in the real scene. However, σ_* needs to be manually adjusted according to the situation of different video sources, and manually adjusting the model parameters is not only inconvenient, but also prone to errors, resulting in a decrease in the accuracy of the algorithm. In order to solve this problem, the parameters are appropriately estimated using the self-organising mechanism [26], that is to say, the constants σ_x and σ_y in Eq. (3), which generate random perturbations, are replaced by the variables σ_x[t] and σ_y[t], which change over time. The augmentation matrix z[t] = [x^T,σ_x,σ_y][t]∈▯⁹ is redefined, and can be obtained by substituting it into the original model: 6 $\begin{matrix} z [t] = [\begin{matrix} A & 0 \\ 0 & I_{2} \end{matrix}] z [t - 1] + [\begin{matrix} v_{x y} \\ v_{c} \\ v_{σ} \end{matrix}] \\ v_{x y} ~ ζ (O, {0, {\hat{σ}}_{x} [t], 0, {\hat{σ}}_{y} [t]}) \in ▯^{4} \\ v_{c} ~ ζ (O, {σ_{c}, σ_{c}, σ_{c}}) \in ▯^{3} \\ v_{σ} ~ ζ (O, {σ_{σ}, σ_{σ}}) \in ▯^{2} \end{matrix}$ 7 $y [t] = C_{A} x z [t]$

In the equation v_xy is a random vector, and v_xy changes as ${\hat{σ}}_{*} [t]$ changes.

Next for constructing the particle swarm equation for the self-organising mechanism, define the set of particle swarms consisting of _M states as: 8 $Z [t] : = [z^{[1]}, z^{[2]}, \dots, z^{[M]}] [t] \in ▯^{9 \times M}$

Above each sampling time point, v_* is randomly generated based on each standard deviation using the winkers algorithm, then prediction particles ${\tilde{z}}^{[m]} [t]$ are computed based on z^[m][t–1], and the difference in degree between each prediction particle and the actual measurements ⁽ⁿ⁾y[t] is computed for the n th detected person via OpenPose: 9 $d^{[m]} = \min_{n} | {}^{(n)}y [t] - C_{A} {\tilde{z}}^{[m]} [t] |, \forall n$

Based on equation (9) the likelihood p^[m] of the m st particle can be calculated and the equation is shown in (10): 10 $p^{[m]} = \frac{1}{\sqrt{2 π α^{2}}} \exp (- \frac{{(d^{[m]})}^{2}}{2 α^{2}})$ where α is used as an adjustable parameter, which is transformed using an exponential function and logarithmic summation in order to avoid numerical computation errors due to very small orders of magnitude. The m th particle can be calculated as: 11 $w^{[m]} = p^{[m]} / \sum_{i = 1}^{M} p^{[i]}$

By comparing the random number (0, 1) with the value of the inverse function f⁻¹(·) of f(·) computed cumulatively by 1 v^[m], the next probabilistic resampling of the new particle is then performed. 12 $\hat{z} [t] = \sum_{i = 1}^{M} w^{[i]} {\tilde{z}}^{[i]} [t] = [{\hat{x}}^{T} {\hat{σ}}_{x} {\hat{σ}}_{y}] [t]$

Where the time series ${\hat{x} [t]}$ corresponds to the tracking information of the target person. Estimates ${\hat{σ}}_{x} [t]$ and ${\hat{σ}}_{y} [t]$ are computed as a self-organising adjustment to obtain v_xy. As known above, the method achieves both estimation of the target state and adjustment of the parameters.

2.1.2

Sparse representation model for human posture

The quantitative description of human posture in physical education sports training is mainly divided into two ways: one is rough posture description [27], the rough description of the described joint parts in a simpler way, such as the position coordinates of the joint points and the distance from other joint points. Its constraints are simpler, but it is easy to lose important information, such as angles. The other is the fine attitude description. The fine attitude description of the limb part of the description of the parameters is rich, including all the joints connected with the joints to form the angle information, the relative relationship between the joint positions, etc.18 The joints of the fine attitude description model can be a good characterisation of the human body posture information, but the fine description of the way will bring a large amount of arithmetic, if all the joints of the fine attitude description will result in the model’s The timeliness of the model will deteriorate. In this paper, we will quantitatively describe the movements of students in sports training and use them as the basis for judgement to calculate the similarity of movements. ${_{x_{i}} i = 1,, N} \subseteq ▯^{n}$ is a set of postures, N is the number of skeletal joints labelled in the training set, and i is the number of skeletal joints. The distance between two joint points is defined as: 13 $d i s (x_{i}, x_{j}) = \sqrt{{(x_{i} - x_{j})}^{T} M (x_{i} - x_{j})}, M \in ▯^{n \times n}$

Since M is a symmetric semipositive definite matrix, it is used to represent the full column rank matrix W with the distance relation between the two joints: 14 $d i s (x_{i}, x_{j}) = \sqrt{{(x_{i} - x_{j})}^{T} W W^{T} (x_{i} - x_{j})}, M \in ▯^{n \times r}, r \leq n$

In this paper, we remove the useless joints such as eyes and ears when characterising sports training movements, and the number of described joints is 14. The two-dimensional coordinates of all the skeletal points are described as a vector of ▯ⁿ, and the skeletal data are sorted according to the serial number, and normalised in rows and columns respectively, to obtain a unified skeletal data format of x_i ∈▯ⁿ,i = 1…N. 1)

Fine posture description method

The fine posture data of the skeletal joint training data numbered i is defined as x_i, and the overall posture description of the human body is expressed as a combination of fine posture data: 15 $X_{j} = \sum_{i = 1}^{N} A_{i j}^{*} x_{i}, A_{j j}^{*} = 0$

The fine attitude representation coefficient matrix $A^{*} = (_{A_{i j}^{*}) N \times N}$ and L₁ optimisation of the coefficient matrix yields angle data: 16 ${_{A^{*}}^{}} = \arg \min | | x_{j} - \sum_{i = 1}^{N} A_{i j} x_{i} | |_{2}^{2} + γ_{a} \sum_{i = 1}^{N} | | A_{i j} | |, \forall j = 1,, N$

Where γ_a > 0 is the penalty coefficient, and by representing X =[x₁,x₂,,,,x_N]∈▯^n×N as a data matrix, A* can be expressed in matrix form: 17 $A^{*} = \arg \min | | X - X A | |_{F}^{2} + γ_{a} | | A | |_{1}$

The final constraints that can be obtained for the fine attitude representation metric are given by Eq: 18 $\begin{array}{l} E_{f u l l} (W, A^{*}) & = {(\sum_{j = 1}^{N} x_{j} - \sum_{i = 1}^{N} A_{i j}^{*} x_{i})}^{T} W W^{T} (x_{j} - \sum_{i = 1}^{N} A_{i j}^{*} x_{i}) \\ = | | {(X - X A^{*})}^{T} W | |_{F}^{2} \end{array}$

2)

Rough Posture Description

The rough posture description model is simpler to describe the human skeleton, and the description of the rough posture data x_j(1:n′) is similar to that of the fine posture, which is described as: 19 $x_{j} (1 : n') = \sum_{i = 1}^{N} B_{i j}^{*} x_{i} (1 : n'), B_{j j}^{*} = 0$

Therefore, the rough attitude representation of the coefficient matrix $B^{*} = (_{B_{\ddot{i} j}^{*}) N \times N}$ is obtained by L₁-optimising the coefficient matrix: 20 $B^{*} = \arg \min ▯ X_{1 : n^{'}} - X_{1 : n^{'}} B ▯_{2}^{2} + γ_{b} ▯ B ▯_{1}$

Where γ_b is the sparse penalty factor and the final rough pose representation metric is: 21 $E_{c o a r s e} (W, B^{*}) = ▯ (^{X_{1 : n^{'}} - X_{1 : n^{'}} | B^{*}) T} W_{1 : n^{'}} ▯_{F}^{2}$

2.2

Unregulated limb labelling methods

In order to help teachers and students in the process of physical education to detect and correct irregular body movements in sports training in a timely manner, this paper implements the function of irregular body marking in sports training. This function is developed and implemented based on the human body movement posture analysis algorithm proposed above, i.e., the movement videos in the sports training to be evaluated and the standard template movement videos are matched and aligned by the human body posture estimation algorithm. Then, the corresponding key action frames are extracted from the motion training actions to be evaluated with reference to the key action frame data in the canonical action database. Finally, the irregular limbs present in these key action frames are labelled.

Let the sequence of human joint angle signatures corresponding to the action video in the exercise training to be evaluated be P = {p₁,p₂,p₃,⋯p_m}, and the sequence of human joint angle signatures corresponding to the standard template action video be Q = {q₁,q₂,q₃,⋯q_n}, wherein p_m is the joint angle signatures vector corresponding to frame m of the action video in the exercise training to be evaluated. q_n the joint angle sign vector corresponding to frame n of the standard template video in the motion training. According to the human posture matching algorithm, the regularisation path, P_m =[θ_p1,θ_p2,⋯,θ_pi,⋯,θ_p8] corresponds to q_n =[θ_q1,θ_q2,⋯,θ_qi,⋯,θ_q8] i.e., for the action video in the exercise training to be evaluated, frame m is aligned with frame n in the standard template action video, and frame n is a pre-calibrated key action frame. The specific steps for irregular limb labelling are as follows. 1)

The deviation vector dif =[dif₁,dif₂,⋯,dif_i,⋯,dif₈] of the joint angles of the action to be evaluated and the action in the motor training is obtained by Equation dif = p_m – q_n.

2)

Define dif_i / θ_qi as the deviation factor of the k th joint angle in the i th frame in the video of the action to be evaluated, and the specification of the limb movement corresponding to this joint angle is measured by D_i. The calculation formula of D_i is shown in Eq: 22 $D_{i} = 1 - \frac{d i f_{i}}{θ_{i}}$

The larger the joint angle deviation degree, the smaller D_i is, and the more irregular the corresponding limb movement is.

3)

The limbs with D_i less than the set value of D₀ are marked, i.e., the coordinates of the key points of the human skeleton calculated by θ_i are used as the reference, and the outer rectangle is drawn as the cue box of the irregular limbs.

2.3

Intelligent sports training management system design

The overall scheme of the intelligent management system combining physical education and sports training is shown in Fig. 1, which can be divided into the visualisation interface module and the GPU auxiliary module in general. The data storage module is primarily responsible for storing user information and standard action data. The GPU analysis and processing module [28] is mainly responsible for segmenting the video of students’ movement actions in sports teaching and sports training through the particle filtering algorithm proposed above in order to achieve a more fine-grained assessment of movement normality. The segmented human skeletal key point coordinate sequences are extracted from the segmented sub-video sequences by the sparse representation model of human posture and returned to the data preprocessing module. Finally, the motion movements are evaluated using the irregular movement identification method. The data preprocessing module is mainly responsible for outlier correction and missing value interpolation of the human skeletal key point sequences extracted by the algorithm and then aligns the action feature sequences by matching and identifying the wrong limbs in the extracted key action frames.

3

Results and discussion

3.1

Test and analysis of human posture estimation methods

In this section, VC2008, combined with the OpenCV image processing library [29], is used to implement human posture estimation based on edge contour features combined with image processing techniques. The steps of testing experiments for the human posture estimation method in physical education and sports training are as follows. 1)

Use the detection algorithm to obtain the binarised contour map of the human body in the image.

2)

Use particle filter edge detection algorithm to extract the contour edges of the binarised contour map.

3)

Use the human body pose sparse representation model proposed in the previous paper to obtain the coordinates of the joints of the human body in the contour edge map. The pose results obtained based on the human pose estimation method proposed in this paper are shown in Fig. 2, with (a)-(c) representing the human contour map, edge contour map, and joint point extraction results, respectively. From the extraction results, it can be found that the human pose estimation algorithm proposed in this paper is able to accurately extract human contours and key points during sports training.

The human joint data obtained based on the method proposed in this paper is now compared with the joint data captured based on Kinect [30] to verify the accuracy of this method. The average relative error is 4.41% according to Table 1, which displays the results of the comparative analysis of joint extraction. Therefore, the pose estimation method based on the particle filtering algorithm and sparse representation model proposed in this paper is more accurate, and its extraction results are close to the theoretical value. This shows that the posture estimation method is able to obtain the posture data of athletes and teachers in the intelligent management system of sports training, and through quantitative analysis and comparison, it can provide intuitive sports analysis guidance for sports teaching.

Table 1.

Comparison of results of node extraction

Closing node	Coordinates		Absolute error (Pixels)	Relative error (%)
Closing node	This method	Kinect extraction	Absolute error (Pixels)	Relative error (%)
Head	(315,369)	(309,357)	13	5.37
Neck	(311,397)	(319,399)	7	4.74
Chest	(309,387)	(304,394)	12	3.05
Left shoulder	(269,348)	(261,346)	10	4.92
Left elbow	(289,415)	(294,420)	12	3.39
Left hand	(419,336)	(415,337)	6	5.04
Right shoulder	(297,365)	(299,358)	13	4.53
Right elbow	(219,364)	(223,374)	11	5.19
Right hand	(359,397)	(344,398)	14	2.67
Left hip	(364,418)	(364,415)	12	2.81
Left knee	(355,59)	(351,58)	6	4
Left foot	(48,265)	(48,264)	10	5.21
Right hip	(129,348)	(129,352)	8	5.41
Right knee	(167,78)	(172,77)	10	5.29
Right foot	(206,198)	(204,198)	15	4.56

3.2

Analysis of the Application Effect of Intelligent Sports Training Management System

This paper applies an intelligent management system that combines physical education and sports training in a sample of schools and analyses three aspects, namely students’ athletic ability, their knowledge of physical education, and teachers’ coaching ability, in order to investigate the extent to which an intelligent training management system affects the quality of teaching and learning.

3.2.1

Sample Selection

A sports school in Wuhan was founded in 1984. As one of the advantageous running majors of the sports school, the sports training major successfully entered the national undergraduate major “Double Ten Thousand Plan” construction point list in 2018 and enjoys a higher reputation and strong strength among similar institutions in the country. Hence, the selection of the sports training major of the school as the evaluation sample of this study is somewhat representative. Therefore, the selection of the sports training program of this university as the assessment sample for this study is representative. Generally speaking, as the main body of the receiving end of the teaching activities, students are the direct beneficiaries of the teaching process and can directly feel and reflect on the quality of teaching. The same class (80 students) was chosen to be divided into two groups, one as the intervention group (38 students), i.e., the intelligent sports training management system was applied to assist in the physical education teaching process. The other group was the non-intervention group (42 students), i.e., continuing the traditional mode of physical education teaching and sports training. The practice period started on May 15, 2023, and ended on September 15, 2023.

3.2.2

Results of analysing students’ motor abilities

1)

Test Indicators

The FMS (Functional Dynamic Screening) is a specialised testing and evaluation tool for basic movement patterns in physical education, a system designed to test the ability of students to perform basic movements, motor control in movement patterns and fundamental movements in sports training using simple movements. The screening test consists of a total of seven movements (deep squat, hurdle step, linear lunge squat, shoulder flexibility, active straight leg raise, trunk stabilisation push-up and rotational stability) and three exclusionary test movements (shoulder impingement clearing test, push-up clearing test and backward swing clearing test).

In this paper, after exporting the students’ exercise data from the intelligent management system, statistics and classification of the data were achieved using SPSS 26.0 software. A database system was constructed using SPSS software, and all the test data were categorised, and their various assessment indices were expressed as mean±standard deviation (M±SD) or median, and the Kolmogorov-Smirnov test was used, and the values of the intervention group and the non-intervention group showed non-normal distribution (P<0.05), and the FMS scores of the intervention group and the non-intervention group were non-normally distributed. Non-parametric tests were employed. The Wilcoxon signed rank sum test was chosen to compare pre-and post-group changes. The Mann-Whitney U test was used to compare the change between groups, and if there was a significant difference in baseline levels between the two groups, the difference in the change in indicators before and after the intervention (diff=post-intervention - pre-intervention) was used for the Mann-Whitney U test. P<0.05 was set as the level of significance.

2)

Analysis of changes in each FMS score

The changes in the FMS indicators before and after the experimental intervention of the two groups of students are shown in Table 2. Comparison within the intervention group revealed that the increase in deep squat (P=0.004), shoulder flexibility (P=0.003), active straight knee raise (P=0.003), trunk stabilisation push-ups (P=0.003), and rotational stability (P=0.001) was significant before and after the experiment (P<0.05), whereas the increase in the two items of over-the-rail racking stride, and the anterior and posterior leg-splitting squat were not significant (P =0.051, 0.063>0.05). Students’ deep squat and shoulder flexibility during physical activity training increased significantly from 2.05 ± 0.05 and 1.75 ± 0.06 to 2.88 ± 0.07 and 2.90 ± 0.06, respectively (W = 1026.22, 1024.41, P < 0.01). Comparisons within the non-intervention group revealed a significant pre- and post-experimental increase in active straight leg raise and rotational stability (P<0.05). An increase in straight knee demonstrated this raises from 1.93±0.06 to 2.44±0.02 (W=1133.08, P=0.041<0.05). It is worth noting that students in the non-intervention group showed a decrease in their scores in the deep squat and over-the-rail racking step after sports training, which indicates that students and teachers in traditional sports training are not able to detect problems in movements in a timely manner, thus leading to a deterioration in the quality of teaching and learning instead.

Table 2.

FMS specific scoring basic situation

Test item	Intervention group (n=38)				Non-intervention group (n=42)
Test item	Before	After	W	P	Before	After	W	P
Squat	2.05±0.05	2.88±0.07	1026.22	0.004	1.95±0.10	1.81±0.07	1323.45	0.059
Hurdle step	1.81±0.04	2.61±0.12	1392.02	0.051	1.96±0.03	1.87±0.11	1637.51	0.097
Squatting	1.98±0.12	2.84±0.05	1221.51	0.063	1.73±0.04	1.98±0.09	1518.67	0.122
Shoulder flexibility	1.75±0.06	2.90±0.06	1024.41	0.003	2.08±0.01	2.41±0.07	1532.86	0.067
Straight knee lift	1.80±0.04	2.36±0.02	1198.26	0.003	1.93±0.06	2.44±0.02	1133.08	0.041
Stable push-ups	2.40±0.03	2.56±0.11	1186.39	0.003	2.19±0.02	2.09±0.05	1525.63	0.089
Rotational stability	1.82±0.08	2.34±0.09	1010.44	0.001	1.75±0.07	2.41±0.01	1195.34	0.045

3.2.3

Analysis of Changes in Sports Perceptions

Sports cognition mainly refers to students’ knowledge and mastery of subjects related to physical education and sports training. Testing students’ sports cognition can demonstrate their mastery of physical education and sports training from a certain perspective, and can assess whether students have achieved the knowledge objectives of the teaching goals. The study mainly tested students’ sports cognition through questionnaires, which were analysed in four aspects: sports knowledge cognition, sports item cognition, sports movement cognition and sports health knowledge cognition.

The results of the homogeneity analysis of the students in the intervention and non-intervention groups in terms of sports cognition are shown in Figure 3, where T1-T4 represents sports knowledge cognition, sports programme cognition, sports movement cognition and sports health knowledge cognition, respectively. The results showed that there was no significant difference between the two groups of students in terms of overall sports cognition. The mean scores of students’ sports cognition in the intervention group were 2.93 (T1), 3.04 (T2), 2.91 (T3) and 3.13 (T4), respectively. The mean scores of students in the non-intervention group were 3.07 (T1), 2.81 (T2), 3.10 (T3) and 3.12 (T4). A t-test of the differences between the students in the intervention and non-intervention groups in terms of sports cognition revealed that t=0.529, P>0.05, which is not a significant difference, indicating that there is little difference between the sports cognitive profiles of the students in the intervention and non-intervention groups, as well as homogeneity.

The post-practice test was conducted to analyse the sports cognition of the two groups of students, and the results of the comparative analysis obtained are shown in Table 3. The scores of the four variables of the post-test of sports cognition in the non-intervention group were 3.46±0.91, 3.45±0.67, 3.27±0.74 and 3.21±0.61, respectively. After the test of statistical significance, the P-value of the four variables of sports cognition in this group was greater than 0.05 (P=0.069-0.136), which is not a significant difference, indicating that the students in the non-intervention group did not improve their sports cognition. On the contrary, the results of the analysis of the intervention group, the students in this group after the application of the intelligent sports training management system, showed the scores of the four variables have been improved (4.60-4.90 points), and the significance value of 0.000, with significant differences, indicating that the intervention group of students in sports cognitively have been significantly improved. Due to the experimental process of physical education in the classroom, teachers teach sports knowledge and students learn during the after-school internalization phase. The main thing is that the application of an intelligent sports training management system can assist students in learning sports and sports training related knowledge visually so that the student’s mastery of sports knowledge has been substantially improved.

Table 3.

Physical cognitive results

Test item		T1	T2	T3	T4
Intervention group (n=38)	Before	3.07±0.53	2.81±0.82	3.10±0.92	3.12±0.78
	After	4.78±0.71	4.90±0.55	4.68±0.51	4.60±0.74
	t	19.641	11.842	17.515	11.599
	P	0.000	0.000	0.000	0.000
Non-intervention group (n=42)	Before	2.93±0.71	3.04±0.97	2.91±0.83	3.13±0.97
	After	3.46±0.91	3.45±0.67	3.27±0.74	3.21±0.61
	t	5.140	4.083	2.643	3.207
	P	0.069	0.115	0.097	0.136

3.2.4

Analysis of Teachers’ Capacity to Teach

Teachers are the guides in teaching and the main body of teaching implementation. The effectiveness of physical education teaching is determined by the teaching ability of teachers, and improving their teaching abilities is an important way to improve teaching quality. Therefore, this paper investigates and analyses the teaching ability of teachers in the intervention group in terms of training organisation ability (Z1), analysis and judgement ability (Z2), effective communication ability (Z3), professional learning ability (Z4), instructional design ability (Z5), and instructional implementation ability (Z6). The results of the analyses of teachers’ competence in teaching by physical education teachers and students in the intervention group are shown in Figure 4. From the teachers’ side, 87.13% and 81.01% of the teachers believed that they had good training organization skills, analytical and judgemental skills. The rest of the selection rate of the teachers in terms of teaching competence was between 60% and 80%. The lowest selection rate was 65.75 per cent for having effective communication skills. From the student’s side, the highest selection rate was 76.13% for teachers who were considered to have good instructional design skills. This was followed by having good professional learning skills at 70.54 per cent. From a comprehensive point of view, both teachers and students choose to think that teachers have good training organisation ability and teaching implementation ability of a higher proportion, which indicates that the intelligent sports training management system in assisting students to standardise sports movements and improve skills at the same time, the teachers also have a promotional effect on the improvement of teaching ability.

4

Conclusion

As a major component of physical education teaching, sports training plays a very important role in the formation and development of students’ sports expertise, and in the comprehensive cultivation and improvement of students’ sports skills and abilities. Based on this, this study proposes a human posture estimation and irregular sports action marking method based on a particle filtering algorithm and sparse model and combines the method to design an intelligent management and analysis system that combines physical education teaching and sports training. In order to investigate its impact on the quality of physical education teaching, this paper sets up a comparative test and analyses it in terms of students’ sports movements, sports cognitive situation and teachers’ teaching ability. It has been found that the average relative error between the human joint data obtained by the method in this paper and the joint data captured using Kinect is 4.41%. It shows that the posture estimation method is capable of obtaining posture data from athletes and teachers in the intelligent management system of sports training, which ensures the effective operation of the system. Comparison within the intervention group found that students before and after the application of the system were significant (P<0.05) in deep squat (P=0.004), shoulder flexibility (P=0.003), active straight knee raise (P=0.003), trunk stabilisation push-up (P=0.003), and rotational stability (P=0.001). It was also found that the application of an intelligent sports training management system can assist students in learning sports and sports training related knowledge through visualisation, so the students in the intervention group got a significant improvement in sports cognition, and the teachers’ teaching ability was also improved.

This paper proves that the application of intelligent management analysis systems in physical education can significantly promote the improvement of the quality of physical education teaching, so it is recommended to actively promote the use of intelligent management systems in physical education teaching.

Language:: English

Publication timeframe:: 1 times per year
Journal Subjects:: Life Sciences, Life Sciences, other, Mathematics, Applied Mathematics, General Mathematics, Physics, Physics, other

Journal RSS Feed

Impact of Intelligent Management of Physical Education Teaching and Athletic Training Integration on Improving Teaching Quality

Chen Cheng

Published Online: Feb 05, 2025

Received: Oct 02, 2024

Accepted: Jan 11, 2025

DOI: https://doi.org/10.2478/amns-2025-0052

KeywordsParticle filtering algorithm, Pose estimation, Sparse model, Key action frames, Sports training

© 2025 Chen Cheng, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Keywords
Particle filtering algorithm, Pose estimation, Sparse model, Key action frames, Sports training