Startseite Rehabilitation training method for motor dysfunction based on video stream matching
Artikel Open Access

Rehabilitation training method for motor dysfunction based on video stream matching

  • Yanyan Hong EMAIL logo
Veröffentlicht/Copyright: 30. Juli 2025
Artikel downloaden (PDF)
Veröffentlichen auch Sie bei De Gruyter Brill
Manuskript einreichen Informationen für Autor*innen
Nonlinear Engineering
Aus der Zeitschrift Nonlinear Engineering Band 14 Heft 1

Abstract

The development of artificial intelligence technology has also promoted the development of the field of sports rehabilitation. The existing motion matching assisted sports rehabilitation technology has poor applicability in complex environments and slow response time. A video stream matching based rehabilitation training method for motor dysfunction is developed to address the existing problems in motion matching in the field of sports rehabilitation. The research design method compares the movement posture to determine the patient’s rehabilitation level and provides rehabilitation guidance. At the same time, a keyframe comparison action similarity measurement algorithm and an unmatched keyframe action flow similarity algorithm were designed for different rehabilitation stages of patients to improve user experience. The results show that the key frame comparison action similarity metric algorithm has a matching time of about 40 ms and a matching accuracy of about 0.85. The unmatched key frame action flow similarity algorithm has no obvious advantage in the matching time of key frames, but it has an obvious advantage in the matching accuracy, which is as high as 0.9. The proposed video flow matching algorithm effectively improves the speed of patients’ rehabilitation training and the patients’ evaluation of this algorithm is higher than that of the video flow matching algorithm. And the patients’ evaluation of the algorithm is high.

Keywords: motor function; video stream matching; rehabilitation training; action recognition

1 Introduction

With technological advances, especially in the fields of artificial intelligence and computer vision, rehabilitation medicine has also ushered in new development opportunities [1]. Especially in the field of rehabilitation training for motor dysfunction, traditional methods are gradually combined with advanced technologies to provide patients with more personalized and efficient rehabilitation programs [2]. Accurate assessment and guidance of the patient’s movement is essential in the rehabilitation process [3]. Traditional rehabilitation often relies on the direct observation and guidance of a physiotherapist, which not only requires a high level of professional competence, but also has resource and time constraints [4]. With the development of video stream matching technology, it is possible to capture and analyze the patient’s movements in real time through computer vision algorithms, thus providing more coherent and objective training feedback [5]. The core of this approach lies in the use of advanced image processing techniques such as motion tracking, posture estimation, and behavioral analysis to achieve accurate matching between patient movements and standard movement patterns [6]. Existing research has proposed online human rehabilitation action recognition algorithms based on monocular vision, which have achieved certain accuracy in experiments. However, the stability and real-time performance of these methods in practical applications still need to be improved [7]. Especially in complex backgrounds or varying lighting conditions, the accuracy of action recognition may significantly decrease. And these action matching algorithms may have insufficient generalization ability and adaptability when facing new patients, different rehabilitation actions, or non-standard execution methods [8]. The research aims to explore a novel approach to rehabilitation training based on video stream matching technology that can monitor and analyze patient movement patterns in real time and compare them with predefined rehabilitation movements to achieve more accurate training and feedback.

The innovation of the study is the combination of video stream matching technology with rehabilitation training, which not only helps to achieve a more personalized training program, but can also provide an important auxiliary tool for physical therapists. By monitoring and analyzing the patient’s movements in real time, movement deviations can be detected and training programs can be adjusted, thus improving the efficiency of rehabilitation. In addition, this method also helps to record and analyze the patient’s rehabilitation process, providing data support for long-term rehabilitation. The study will be carried out in four parts, the first part is a review of the current status of research on motor dysfunction and video stream matching, the second part is a study of the method of motor rehabilitation training based on video stream matching, the third part is an analysis of the effect of rehabilitation training for motor dysfunction, and the last part is a summary of the study.

2 Related works

Video stream matching is a process of matching and recognizing objects or features in a continuous video stream using computer vision and image processing techniques. Ye et al. proposed a dynamic graph matching framework in order to solve the problem of cross-camera label estimation. The framework iteratively improves the graph structure by learning similarity metrics, thus improving the label estimation process. Experimental results show that the DGM outperforms state-of-the-art unsupervised re-ID methods, yielding performance comparable to a fully supervised upper bound [9]. Fried O et al. proposed a novel method in order to edit the head video of a person who is talking. The method annotates parameters seamlessly stitched to produce an intermediate representation and converts to real video using a loop video generation network. The results show that the method works well in various editing, language translation and sentence synthesis [10]. Sreenu and Durai, in order to solve the problem of violence detection in security surveillance videos in unstructured big data, proposed an approach based on deep learning techniques to improve video target matching. The recognition problems and future directions of the existing methods are summarized by comparing the underlying deep learning implementation techniques and real-time processing considerations [11]. Jabnoun et al., in order to help blind people better interact with their surroundings, proposed a visual aid tool based on the recognition of objects in a video scene. The tool achieves recognition of objects in different frames by calculating the differences between frames and optimizing video processing. The results show that the method can be an option for solving the problem of visual aids for blind and disabled people [12]. Kittel et al., in order to investigate the effectiveness of 360° virtual reality (VR) and match broadcast footage in improving decision making, conducted a randomized controlled trial with Australian amateur soccer referees. The results showed that the 360° VR group performed significantly better than the control group in a retention trial and scored higher in terms of psychological realism, enjoyment, and relevance [13].

Rehabilitation training for motor dysfunction can help patients with disabilities, or mobility problems due to illness, to return to normal living conditions as early as possible. Xu et al., in order to solve the problem of hand and bone motor dysfunction, proposed a new soft hand exoskeleton robot to optimize the force transfer path and improve the local structure. The results show that the exoskeleton can provide stable gripping force to meet the needs of daily life, and the effectiveness of the assisted grasping method is successfully verified through experiments [14]. Li et al., in order to solve the motor and sensory dysfunction and central facial paralysis problems of stroke patients, proposed a rehabilitation training method based on a VR game and combined with a haptic feedback device to improve the patient’s participation and training effect. The experimental results show that the method can reduce trajectory deviation and improve user performance, which is expected to attract patients to do more training [15]. Xiong et al. explored the effect of head acupuncture combined with cognitive training on the cognitive and motor functions of recovering stroke patients, wherein 70 patients were divided into the experimental group and the control group. The results showed that the mini-mental state examination, Loewenstein occupational therapy cognitive assessment, and Fugl-Meyer assessment scores of the experimental group were significantly higher than those of the control group at 12 weeks after treatment, and the plasma BDNF and NGF levels were also significantly increased [16]. Shi et al., in order to help the patients with damaged lower limbs to recover the normal walking ability, proposed a lower limb rehabilitation exoskeleton robot. The results showed that the robot can simulate normal gait and realize robot-assisted rehabilitation training by means of bionics, robotics, information and control science, medicine, and other technologies [17].

According to the investigation of existing research, common rehabilitation training methods for motor dysfunction, including all kinds of robot assistance, rehabilitation training guidance from healthcare personnel, or neural stimulation methods, video stream matching can effectively extract the target’s movements in the video and compare the target movements with the reference movements, so as to calibrate the completion of the target movements, rehabilitation training for motor dysfunction, mostly all kinds of patients on the various types of rehabilitation movement training completion. Therefore, the study proposes to use the video stream matching method to identify and match the rehabilitation actions of patients with motor dysfunction, in order to analyze the completion of the patients’ actions, and to help the patients to complete the rehabilitation training more efficiently.

3 Research on sports rehabilitation training method based on video stream matching

Patients with motor function have many inconveniences in daily life, and motor rehabilitation training can help patients restore basic motor ability and help them return to normal life. The study introduces a method of rehabilitation training for motor dysfunction based on video stream matching in this chapter, which is divided into two parts, the first part is the study of human posture recognition algorithm, and the second part is the study of rehabilitation training based on video stream matching.

3.1 Research on human posture recognition model based on MoveNet model

When patients with motor dysfunction undergo rehabilitation training, they need to complete a variety of specific rehabilitation movements, and the completion of these rehabilitation movements can reflect the recovery status of the patient’s motor function [18]. MoveNet is a lightweight and efficient real-time human posture estimation model [19]. It can realize real-time detection and tracking of human posture, and can accurately identify the key parts of the human body and track the movement of these parts over time. The model performs feature extraction of task actions with mobilieNetV2. The mobilieNetV2 network structure is shown in Figure 1.

Figure 1 MobileNetV2 network structure.
Figure 1

MobileNetV2 network structure.

OpenPose is a model for human pose estimation and multi-person key point detection, which uses deep learning techniques for human key point detection, and can detect and track body poses of multiple people in real time [20]. It not only detects human key points such as head, shoulder, elbow, wrist, hip, knee, etc. but also infers the connections between these key points to accurately localize human movements in images or videos [21]. It has a wide range of application potential for human pose estimation and multi-person keypoint detection, and is widely welcomed by the scientific research and industrial communities [22]. Both MoveNet and OpenPose are commonly used algorithms for human action recognition, which need to extract the features of the human action, MoveNet with mobilieNetV2 structure, for the extraction of the motion gesture features, and the OpenPose models’ BackBone VGGNet structure is used for feature extraction. In the character action recognition algorithm, the OpenPose model has better feature extraction effect and the MoveNet model has better operation effect. Therefore, the research combines the two and replaces the feature extraction module of the MoveNet model with BackBone VGGNet, which improves the feature extraction accuracy of the MoveNet model while retaining the adaptability of its framework to the system, and the research constructs a pose recognition network model as shown in Figure 2.

Figure 2 VGGNet MoveNet pose recognition model.
Figure 2

VGGNet MoveNet pose recognition model.

The model constructed by the study needs to perform the operations of character key point prediction, key point heat map, center heat map, and local offset simultaneously after completing the task action feature extraction. The study’s improvement of the gesture recognition model mainly focuses on the optimization of the feature extraction network. The feature extraction network of the VGGNet-MoveNet model is VGGNet, which is divided into six types, and the study focuses on the VGG-16 network to optimize it. Compared to mobileNetV2, VGG-16 has a higher consumption of system resources, which causes this network to be unsuitable in MoveNet. The study optimizes VGG-16 by changing the input size of the network and replacing the fully connected part of the network with a bottleneck of the network consisting of a convolutional kernel, and the optimized network presents a fully convolutional structure, as shown in Figure 3.

Figure 3 Optimizing the VGG-16 network structure.
Figure 3

Optimizing the VGG-16 network structure.

After completing the optimization of the pose recognition network in the study, it is necessary to construct the loss function of the network, which is a pose recognition network, and therefore, the loss function will be affected by the Header features of the network. The loss function used when key point prediction and local offset are used as Header features is shown in Eq. (1).

(1) L 1 = ∑ i n ∣ Y i ′ − Y i ∣ ,

where Y i denotes the feature true label and Y i ′ denotes the feature predicted label. The loss function used when using the center heatmap and keypoint heatmap as Header features is shown in Eq. (2).

(2) L 2 = ∑ i n ( Y i ′ − Y i ) 2 .

The loss function of key point prediction, local offset, center heat map, and key point heat map can be calculated by Eqs. (1) and (2). According to this feature loss function, the Google key point position of the target task in the image can be calculated, and at the same time, the offset of the skeleton can be counted into the loss as an evaluation result. Therefore, the final loss function of this network is the sum of five feature losses, as shown in Eq. (3).

(3) L total = L reg + L offset + L ctr + L kh + L bk ,

where L total denotes the total loss function; L reg denotes the good key point prediction loss function; L offset denotes the local offset loss function; L ctr denotes the center heat map loss function; L kh denotes the key point heat map loss function; and L bk denotes the skeletal offset loss. The error in pose recognition is also a point that the model must consider. In the training process of the network, an error threshold should be set, and the error sum of each key point needs to be less than the error threshold, as shown in Eq. (4).

(4) E = ∑ k = 1 K e k < E thre = 2 K ⋅ E pix ,

where E thre denotes the error threshold, e k denotes the key point error, and E denotes the sum of key point errors. The accuracy of the model is also one of the necessary indicators, and the calculation of the accuracy is shown in Eq. (5).

(5) P ACC = N accurate N total ,

where P ACC denotes the model accuracy; N accurate denotes the number of coordinate points that meet the requirements, and N total denotes the total number of coordinate points. The model designed in the study is a multi-person pose recognition network, which needs to calculate the similarity of the target key when facing multi-person pose recognition, as shown in Eq. (6).

(6) OKS = ∑ i N e − d i 2 2 S 2 σ i 2 ⋅ δ ( v i > 0 ) / ∑ i N δ ( v i > 0 ) ,

where OKS denotes the similarity of keypoints; N denotes the number of keypoints; d i denotes the Euclidean distance; S denotes the scale feature of the main task; v denotes the visibility of the keypoints of the main task; σ denotes the normalization factor; and δ denotes the unit function. The average accuracy is an important index in multi-person pose recognition, and its calculation is shown in Eq. (7).

(7) AP = Σ p δ ( OKS P > T ) p ,

where p denotes the number of consecutive detections. The number of transmitted frames per second is also one of the important indicators when recognizing video content, and the formula for this indicator is shown in Eq. (8).

(8) FPS = 1 T process ,

where FPS denotes the number of transmitted frames per second; T process denotes the processing time of each frame. It is the sum of multiple times as shown in Eq. (9).

(9) T pre + T NN + T post ,

where T pre represents the preprocessing time of the screen; T NN represents the consumption time of the network prediction; and T post represents the post-processing time of the image. After the study of the optimization of the algorithm, the use of the core of the front and back operations is different. At this time, T process will be determined by the bottleneck in the image processing process, as shown in Eq. (10).

(10) T process = Max ( T pre , T NN , T post ) .

3.2 Research on rehabilitation training based on video stream matching

Rehabilitation training based on video stream matching requires similarity matching of user’s actions and postures. Similarity comparison includes single-frame similarity metric and video stream similarity metric. Single-frame similarity metric needs to select the key points of the character’s movements. In the study in the single-frame similarity metric, 12 joints and 5 key parts of the human body’s movements are selected as the key points of the similarity metric, as shown in Figure 4.

Figure 4 Distribution of key points for action similarity measurement.
Figure 4

Distribution of key points for action similarity measurement.

The uniaxial similarity metric is divided into three parts: calculating the Euclidean distance after normalization, calculating the Euclidean distance for joint angle vectors, and comparison learning. In human posture estimation, quantifying the movement differences by calculating the Euclidean distance of key point vectors is key. However, the raw data may cause errors due to different relative positions of the user and the camera. For this reason, normalization techniques, such as max-min normalization or z-score normalization, are used to adjust the data scale to ensure that all key points are on the same scale of proportions. Although this method reduces the shooting angle and distance errors, it cannot solve the problem of the difference in body proportions between the user and the demonstrator. Therefore, in calculating the Euclidean distance to solve the user-demonstrator stature scale difference problem through the joint angle vector, the normalization of the horizontal coordinates of the keypoints is shown in Eq. (11).

(11) x ′ = ( x − x min ) / ( x max − x min ) ,

where x denotes the horizontal coordinates of the key point, x max , x min denotes the extreme value of the horizontal coordinates, and x ' denotes the normalized horizontal coordinates. The normalization of the vertical coordinates of the key point is shown in Eq. (12).

(12) y ′ = ( y − y min ) / ( y max − y min ) ,

where y denotes the key point longitudinal coordinate, y max , y min denotes the longitudinal coordinate extreme value; y ′ denotes the normalized longitudinal coordinate. When studying the recognition of character poses, random sampling is adopted, which requires setting a random parameter to extract action poses. The extraction of actions follows a Gaussian distribution of absolute values, and the probability density function ϕ ( x ) can be expressed as (13) [23].

(13) ϕ ( x ) = 2 π π Σ e − x 2 2 Σ 2 ( x > 0 ) 0 ( x ≤ 0 ) .

At this point, the calculated probability can be expressed as Eq. (14) [24].

(14) P x < X = ∫ 0 X ϕ ( x ) d x .

After the calculation of the above two Euclidean distances, the user actions can be compared and learned with the demonstrator’s actions. Comparison learning creates a comparison learning dataset by taking the current frame and its neighboring frames as the positive samples for comparison learning during the learning process, while considering the current frame and the frames that are farther away in the time sequence as the negative samples for comparison learning of the group. The video stream similarity metric calculates the similarity of two action streams based on the single-frame similarity metric. In the early stage of rehabilitation training, all kinds of muscle exercise and stretching movements are used, and there is usually a key frame in these movements, so by comparing the key frame in the demonstrator’s movements and the patient’s movements, it can be effectively judged whether the patient’s movement posture meets the standard requirements. Therefore, the study proposes an action similarity measurement for keyframe comparison (ASMKC) algorithm, which is divided into two phases, the first phase is the keyframe extraction of the user and the demonstrator’s actions, and the second part is the similarity matching of the action flow. The specific flow is shown in Figure 5.

Figure 5 ASMKC algorithm process.
Figure 5

ASMKC algorithm process.

In the algorithm shown in Figure 5, action similarity matching, is the core part of the algorithm. When the action flow similarity matching, the key frame sequence of the demonstrator’s action needs to be sorted and compared with the user’s action one by one, if the distance between the key frame of the demonstrator’s action and the user’s current frame is less than a set threshold, it can be regarded as a standard action, and vice versa if the action is not up to the standard, until the complete comparison of all the key frames of the action. If the patient’s rehabilitation training has entered the middle stage, the patient can gradually accelerate the training speed of the rehabilitation action, then the ASMKC algorithm will have the situation that the user’s action is not detected. Therefore, the study proposes an unmatched keyframe action flow similarity (UKAFS) algorithm, which can obtain the ASMKC algorithm, the keyframe actions that are not matched will be extracted for the keyframe sequence matching and be run to skip the keyframes and match with the later sequence frames for matching, the flow of this algorithm is shown in Figure 6.

Figure 6 UKAFS algorithm process.
Figure 6

UKAFS algorithm process.

4 Analysis of the effect of rehabilitation training for motor function disorders

The study constructs a VGGNet-MoveNet model in Chapter 3 and proposes a rehabilitation training method for motor dysfunction based on video stream matching based on this model. The main content of this chapter is the analysis of the effect of the method in practical application, which is divided into two parts. The first part is the setting of the experimental environment parameters, and the second part is the validation analysis of the model, algorithm, and the effect of the rehabilitation training method.

4.1 Experimental environment and parameter settings

In the simulation experimental analysis of the model, the device operating system used is windows 7 64 bit, the processor is Intel(R) Core (TM) i5-4460 CPU@3.20 GHz 3.20 GHz, the system memory is 16GB, the deep learning framework is Pytorch, the computer vision library is OpenCV4, the algorithmic programming language is C++. The training data for the model is the COCO dataset, which contains 80 kinds of recognized objects with a total of 200,000 images, and the parameter settings of the VGGNet-MoveNet model are shown in Table 1.

Table 1

Model parameter settings

Parameter Value
Learn rate 0.001
Weight decay 0.0001
Batch size 32
Input size 192*192*3
Epoch 500

4.2 Analysis of the effect of rehabilitation training

The VGGNet-MoveNet model designed by the study was obtained by optimizing the VGGNet structure based on the MoveNet model. Therefore, the study compared the VGGNet-MoveNet model with the MoveNet model loss degree based on the root mean square error (RMSE). The results are shown in Figure 7.

Figure 7 Model loss degree comparison results.
Figure 7

Model loss degree comparison results.

As can be seen in Figure 7, with the increase in the number of iterations, the loss degree of both models is decreasing rapidly. After 100 iterations, the MoveNet model completes the convergence, and the RMSE value of the converged MoveNet model is around 0.25, and the MoveNet model begins to converge for the first time at around 25 iterations, and then completes the convergence at the 100th iteration, and the VGGNet-MoveNet model is prone to fall into local optimality during training. The MoveNet model is prone to fall into local optimum during the training process. The VGGNet-MoveNet model completes its convergence at the 150th iteration, and the RMSE value of the converged VGGNet-MoveNet model is around 0.10. The VGGNet-MoveNet model converges with a smaller degree of loss, and the MoveNet model converges more quickly. The study also analyzed the action recognition accuracy of the two models and the accuracy of the models in comparison and the results are shown in Figure 8.

Figure 8 Comparison results of accuracy and relative accuracy. (a) Accuracy comparison results. (b) Comparison results of relative accuracy.
Figure 8

Comparison results of accuracy and relative accuracy. (a) Accuracy comparison results. (b) Comparison results of relative accuracy.

Figure 8(a) shows the results of the model’s accuracy with the increase in the number of samples, and it can be seen that with the increase in the number of training samples, the model’s accuracy shows a trend of increasing first and then regionally smooth. When the number of samples is 20, the accuracy of the MoveNet model starts to level off, and at this time, the accuracy of the model is around 84%. When the number of samples is 10, the accuracy of the VGGNet-MoveNet model starts to level off, and at this time, the accuracy of the model is around 93%. Figure 8(b) shows the results of the model accuracy with the number of iterations, and in the first 50 iterations of training, the accuracy of both models increased rapidly, with the highest accuracy of about 0.82 for the VGGNet-MoveNet model, and the highest accuracy of about 0.78 for the MoveNet model. After completing the training of the models, the study compares the recognition efficiency and recognition accuracy of the video stream matching algorithms, and the results of the comparison between the ASMKC algorithm and the methods in the literature [21] are shown in Figure 9.

Figure 9 The algorithm performance comparison results. (a) Time consumption comparison. (b) Comparison results of matching accuracy.
Figure 9

The algorithm performance comparison results. (a) Time consumption comparison. (b) Comparison results of matching accuracy.

Figure 9(a) shows the comparison of the two algorithms’ time consumption for early action keyframe matching, and it can be seen that the ASMKC algorithm’s time consumption for single action keyframe matching is always lower than that of the method in literature [21]. The ASMKC algorithm’s time consumption for single action keyframe matching is the highest at 42 ms, while that of the method in literature [19] is the lowest at 96 ms for single action keyframe matching. Figure 9(b) shows the comparison of the accuracy of the two algorithms for early action keyframe matching, and it can be seen that in individual samples, the method in the literature [21] has a higher recognition accuracy. In action 4, the recognition accuracy of the method in literature [21] is higher, while in the rest of the samples, the ASKMC algorithm has a higher recognition accuracy. The results of the comparison between UKAFS and the method in literature [22] are shown in Figure 10.

Figure 10 The algorithm performance comparison results. (a) Time consumption comparison. (b) Comparison results of matching accuracy.
Figure 10

The algorithm performance comparison results. (a) Time consumption comparison. (b) Comparison results of matching accuracy.

Figure 10(a) shows a comparison of the time required for two algorithms to complete mid training action key frame matching, and it can be seen that, in the middle of the patient’s action is completed at a faster speed, the time consumed by the two algorithms to complete the matching of the key frames of the single action is relatively close to the two algorithms, and the two algorithms do not have any obvious advantage. Figure 10(b) shows the comparison of the accuracy of the two algorithms in matching the keyframes of a single action in the mid-term training action, and it can be seen that the accuracy of the UKAFS algorithm in matching the keyframes of the action is much higher than that of the method in the literature [22]. The keyframe matching accuracy of the UKAFS algorithm in the fast action can reach more than 0.9, whereas the highest matching accuracy of the method in the literature [22] is only about 0.73. The study concludes by comparing the rehabilitation training method proposed in the article with the traditional rehabilitation training method, and the results are shown in Figure 11.

Figure 11 Comparison of rehabilitation training effects. (a) Time consumption comparison. (b) Comparison of ratings.
Figure 11

Comparison of rehabilitation training effects. (a) Time consumption comparison. (b) Comparison of ratings.

Figure 11(a) shows the comparison of the time for the two methods to complete the patient’s rehabilitation training, and it can be seen that the method proposed by the study can always help the patient to complete the rehabilitation training in a shorter period of time compared to the traditional methods of motor dysfunction response. Most of the traditional methods complete the rehabilitation training in about 45 days, while the method proposed by this study completes the rehabilitation training in about 30 days. The proposed method can effectively improve patients’ completion of rehabilitation movements, help patient’s complete rehabilitation training faster, and restore basic motor functions earlier. The study issued a questionnaire to the test volunteers and counted the volunteers’ scores on the two methods, as shown in Figure 11(b). It can be seen that the highest score of the method proposed by the study is 95 and the lowest is 87, while the highest score of the traditional method is 84 and the lowest score is only 68. The method proposed by the study can help the patients to complete the rehabilitation training more quickly and the patients rated the method proposed by the study more highly.

5 Conclusion

In order to improve the efficiency of rehabilitation training for patients with motor dysfunction, the study proposes a video stream matching based rehabilitation training method for motor dysfunction. This is based on the VGGNet-MoveNet model, which is used to recognize the patient’s postures, and the video stream matching of the patient’s rehabilitation movements through the ASMKC algorithm and the UKAFS algorithm, where the ASMKC algorithm is used for the early patient’s rehabilitation movement matching and UKAFS algorithm is used for mid-term rehabilitation training movement matching of the patient. The results show that the loss degree of the VGGNet-MoveNet model after convergence is around 0.10, the highest recognition accuracy is around 0.82, and the model recognition accuracy is around 93%. The ASMKC algorithm takes about 40 ms to complete the matching of a single action keyframe, while the method in the study by Kakavas et al. [21] takes about 100 ms to complete the matching.

In summary, the main conclusions of the study are as follows: First, compared to the MoveNet model, the VGGNet-MoveNet model has lower losses and higher pose recognition accuracy. Second, the ASMKC algorithm outperforms existing methods in both the matching time and accuracy of rehabilitation action keyframes. Third, the UKAFS algorithm does not have a significant advantage in completing single action keyframe matching, but its matching accuracy for keyframes is significantly higher than existing methods.

The research proposed, video stream matching based rehabilitation training method for motor dysfunction, can effectively improve the elapsed time for patients to complete rehabilitation training. The proposed method, by improving the completion of the patient’s rehabilitation movements, improves the efficiency of the patient’s rehabilitation training, and optimizes the movements of the patients’ rehabilitation training. Also, the method can be embarked on in the future to improve the training movements of motor dysfunction from the patient’s rehabilitation movements, and to improve the efficiency of the patient’s rehabilitation training for motor dysfunction.

  1. Funding information: Authors state no funding involved.

  2. Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.

  3. Conflict of interest: Authors state no conflict of interest.

  4. Data availability statement: All data generated or analyzed during this study are included in this published article.

References

[1] Della Villa F, Buckthorpe M, Grassi A, Nabiuzzi A, Tosarelli F, Zaffagnini S, et al. Systematic video analysis of ACL injuries in professional male football (soccer): Injury mechanisms, situational patterns and biomechanics study on 134 consecutive cases. Br J Sports Med. 2020;54(23):1423–32.10.1136/bjsports-2019-101247Suche in Google Scholar PubMed

[2] Jiao L, Zhang R, Liu F, Yang S, Hou B, Li L, et al. New generation deep learning for video object detection: a survey. IEEE Trans Neural Netw Learn Syst. 2021;33(8):3195–215.10.1109/TNNLS.2021.3053249Suche in Google Scholar PubMed

[3] Lyu D, Wang J, Yang F, Zhang Y, Zhang W, Liu H, et al. Effect of Tai Chi on post-stroke non-motor disorders: a systematic review and meta-analysis of randomized controlled trials. Clin Rehabil. 2021;35(1):26–38.10.1177/0269215520951020Suche in Google Scholar PubMed

[4] Bao X, Li L, Liu H, Shi HF, Xu S, Wang MH, et al. Effect of acupuncture combined with rehabilitation on cognitive and motor functions in poststroke patients. Anat Rec. 2021;304(11):2531–7.10.1002/ar.24700Suche in Google Scholar PubMed

[5] Lu K, Xu G, Li W, Huo C, Liu Q, Lv Z, et al. Frequency-specific functional connectivity related to the rehabilitation task of stroke patients. Med Phys. 2019;46(4):1545–60.10.1002/mp.13398Suche in Google Scholar PubMed

[6] Eschweiler M, Bohr L, Kessler J, Fink GR, Kalbe E, Onur OA. Combined cognitive and motor training improves the outcome in the early phase after stroke and prevents a decline of executive functions: a pilot study. Neuro Rehabil. 2021;48(1):97–108.10.3233/NRE-201583Suche in Google Scholar PubMed

[7] Wang C, Feng Y, Zhong L, Zhu S, Zhang C, Zheng S, et al. UbiPhysio: Support daily functioning, fitness, and rehabilitation with action understanding and feedback in natural language. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies. Vol. 8, No. 1, 2024. p. 1–27.10.1145/3643552Suche in Google Scholar

[8] Biswas A, Rao PD, Madhavan S, Natarajan M, Solomon JM. Video parameters for action observation training in stroke rehabilitation: a scoping review. Disability Rehabil. 2024;46(7):1256–65.10.1080/09638288.2023.2191016Suche in Google Scholar PubMed

[9] Ye M, Li J, Ma AJ, Zheng L, Yuen PC. Dynamic graph co-matching for unsupervised video-based person re-identification. IEEE Trans Image Process. 2019;28(6):2976–90.10.1109/TIP.2019.2893066Suche in Google Scholar PubMed

[10] Fried O, Tewari A, Zollhöfer M, Finkelstein A, Shechtman E, Goldman DB, et al. Text-based editing of talking-head video. ACM Trans Graph (TOG). 2019;38(4):1–14.10.1145/3306346.3323028Suche in Google Scholar

[11] Sreenu G, Durai S. Intelligent video surveillance: A review through deep learning techniques for crowd analysis. J Big Data. 2019;6(1):1–27.10.1186/s40537-019-0212-5Suche in Google Scholar

[12] Jabnoun H, Benzarti F, Morain-Nicolier F, Amiri H. Video-based assistive aid for blind people using object recognition in dissimilar frames. Int J Adv Intell Paradig. 2019;14(1–2):122–39.10.1504/IJAIP.2019.102967Suche in Google Scholar

[13] Kittel A, Larkin P, Elsworthy N, Lindsay R, Spittle M. Effectiveness of 360 virtual reality and match broadcast video to improve decision-making skill. Sci Med Footb. 2020;4(4):255–62.10.1080/24733938.2020.1754449Suche in Google Scholar

[14] Xu D, Wu Q, Zhu Y. Development of a soft cable-driven hand exoskeleton for assisted rehabilitation training. Ind Rob: Int J Robot Res Appl. 2021;48(2):189–98.10.1108/IR-06-2020-0127Suche in Google Scholar

[15] Li M, Wu Z, Zhao CG, Yuan H, Wang T, Xie J, et al. Facial expressions-controlled flight game with haptic feedback for stroke rehabilitation: A proof-of -concept study. IEEE Robot Autom Lett. 2022;7(3):6351–8.10.1109/LRA.2022.3170214Suche in Google Scholar

[16] Xiong J, Zhang Z, Ma Y, Li Z, Zhou F, Qiao N, et al. The effect of combined scalp acupuncture and cognitive training in patients with stroke on cognitive and motor functions. Neuro Rehabil. 2020;46(1):75–82.10.3233/NRE-192942Suche in Google Scholar PubMed

[17] Shi D, Zhang W, Zhang W, Ding X. A review on lower limb rehabilitation exoskeleton robots. Chin J Mech Eng. 2019;32(1):1–11.10.1186/s10033-019-0389-8Suche in Google Scholar

[18] Capecci M, Ceravolo MG, Ferracuti F, Iarlori S, Monteriu A, Romeo L, et al. The KIMORE dataset: KInematic assessment of MOvement and clinical scores for remote monitoring of physical REhabilitation. IEEE Trans Neural Syst Rehabil Eng. 2019;27(7):1436–48.10.1109/TNSRE.2019.2923060Suche in Google Scholar PubMed

[19] Li J, Wang Z, Qiu S, Zhao H, Wang Q, Plettemeier D, et al. Using body sensor network to measure the effect of rehabilitation therapy on improvement of lower limb motor function in children with spastic diplegia. IEEE Trans Instrum Meas. 2020;69(11):9215–27.10.1109/TIM.2020.2997545Suche in Google Scholar

[20] Mokayed H, Quan TZ, Alkhaled L, Sivakumar V. Real-time human detection and counting system using deep learning computer vision techniques. Artif Intell Appl. 2023;1(4):221–9.10.47852/bonviewAIA2202391Suche in Google Scholar

[21] Kakavas G, Malliaropoulos N, Bikos G, Pruna R, Valle X, Tsaklis P, et al. Periodization in anterior cruciate ligament rehabilitation: A novel framework. Med Princ Pract. 2021;30(2):101–8.10.1159/000511228Suche in Google Scholar PubMed PubMed Central

[22] Barton JJS, Ranalli PJ. Vision therapy: Ocular motor training in mild traumatic brain injury. Ann Neurol. 2020;88(3):453–61.10.1002/ana.25820Suche in Google Scholar PubMed

[23] Zhang B, Chen J, Xu Y, Zhang H, Yang X, Geng X. Auto-encoding score distribution regression for action quality assessment. Neural Comput Appl. 2024;36(2):929–42.10.1007/s00521-023-09068-wSuche in Google Scholar

[24] Khan MA, Javed K, Khan SA, Saba T, Habib U, Khan JA, et al. Human action recognition using fusion of multiview and deep features: An application to video surveillance. Multimed Tools Appl. 2024;83(5):14885–911.10.1007/s11042-020-08806-9Suche in Google Scholar
Received: 2024-09-05
Revised: 2024-10-21
Accepted: 2024-12-11
Published Online: 2025-07-30

© 2025 the author(s), published by De Gruyter

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

Artikel in diesem Heft

  1. Research Articles
  2. Generalized (ψ,φ)-contraction to investigate Volterra integral inclusions and fractal fractional PDEs in super-metric space with numerical experiments
  3. Solitons in ultrasound imaging: Exploring applications and enhancements via the Westervelt equation
  4. Stochastic improved Simpson for solving nonlinear fractional-order systems using product integration rules
  5. Exploring dynamical features like bifurcation assessment, sensitivity visualization, and solitary wave solutions of the integrable Akbota equation
  6. Research on surface defect detection method and optimization of paper-plastic composite bag based on improved combined segmentation algorithm
  7. Impact the sulphur content in Iraqi crude oil on the mechanical properties and corrosion behaviour of carbon steel in various types of API 5L pipelines and ASTM 106 grade B
  8. Unravelling quiescent optical solitons: An exploration of the complex Ginzburg–Landau equation with nonlinear chromatic dispersion and self-phase modulation
  9. Perturbation-iteration approach for fractional-order logistic differential equations
  10. Variational formulations for the Euler and Navier–Stokes systems in fluid mechanics and related models
  11. Rotor response to unbalanced load and system performance considering variable bearing profile
  12. DeepFowl: Disease prediction from chicken excreta images using deep learning
  13. Channel flow of Ellis fluid due to cilia motion
  14. A case study of fractional-order varicella virus model to nonlinear dynamics strategy for control and prevalence
  15. Multi-point estimation weldment recognition and estimation of pose with data-driven robotics design
  16. Analysis of Hall current and nonuniform heating effects on magneto-convection between vertically aligned plates under the influence of electric and magnetic fields
  17. A comparative study on residual power series method and differential transform method through the time-fractional telegraph equation
  18. Insights from the nonlinear Schrödinger–Hirota equation with chromatic dispersion: Dynamics in fiber–optic communication
  19. Mathematical analysis of Jeffrey ferrofluid on stretching surface with the Darcy–Forchheimer model
  20. Exploring the interaction between lump, stripe and double-stripe, and periodic wave solutions of the Konopelchenko–Dubrovsky–Kaup–Kupershmidt system
  21. Computational investigation of tuberculosis and HIV/AIDS co-infection in fuzzy environment
  22. Signature verification by geometry and image processing
  23. Theoretical and numerical approach for quantifying sensitivity to system parameters of nonlinear systems
  24. Chaotic behaviors, stability, and solitary wave propagations of M-fractional LWE equation in magneto-electro-elastic circular rod
  25. Dynamic analysis and optimization of syphilis spread: Simulations, integrating treatment and public health interventions
  26. Visco-thermoelastic rectangular plate under uniform loading: A study of deflection
  27. Threshold dynamics and optimal control of an epidemiological smoking model
  28. Numerical computational model for an unsteady hybrid nanofluid flow in a porous medium past an MHD rotating sheet
  29. Regression prediction model of fabric brightness based on light and shadow reconstruction of layered images
  30. Dynamics and prevention of gemini virus infection in red chili crops studied with generalized fractional operator: Analysis and modeling
  31. Qualitative analysis on existence and stability of nonlinear fractional dynamic equations on time scales
  32. Fractional-order super-twisting sliding mode active disturbance rejection control for electro-hydraulic position servo systems
  33. Analytical exploration and parametric insights into optical solitons in magneto-optic waveguides: Advances in nonlinear dynamics for applied sciences
  34. Bifurcation dynamics and optical soliton structures in the nonlinear Schrödinger–Bopp–Podolsky system
  35. Review Article
  36. Haar wavelet collocation method for existence and numerical solutions of fourth-order integro-differential equations with bounded coefficients
  37. Special Issue: Nonlinear Analysis and Design of Communication Networks for IoT Applications - Part II
  38. Silicon-based all-optical wavelength converter for on-chip optical interconnection
  39. Research on a path-tracking control system of unmanned rollers based on an optimization algorithm and real-time feedback
  40. Analysis of the sports action recognition model based on the LSTM recurrent neural network
  41. Industrial robot trajectory error compensation based on enhanced transfer convolutional neural networks
  42. Research on IoT network performance prediction model of power grid warehouse based on nonlinear GA-BP neural network
  43. Interactive recommendation of social network communication between cities based on GNN and user preferences
  44. Application of improved P-BEM in time varying channel prediction in 5G high-speed mobile communication system
  45. Construction of a BIM smart building collaborative design model combining the Internet of Things
  46. Optimizing malicious website prediction: An advanced XGBoost-based machine learning model
  47. Economic operation analysis of the power grid combining communication network and distributed optimization algorithm
  48. Sports video temporal action detection technology based on an improved MSST algorithm
  49. Internet of things data security and privacy protection based on improved federated learning
  50. Enterprise power emission reduction technology based on the LSTM–SVM model
  51. Construction of multi-style face models based on artistic image generation algorithms
  52. Research and application of interactive digital twin monitoring system for photovoltaic power station based on global perception
  53. Special Issue: Decision and Control in Nonlinear Systems - Part II
  54. Animation video frame prediction based on ConvGRU fine-grained synthesis flow
  55. Application of GGNN inference propagation model for martial art intensity evaluation
  56. Benefit evaluation of building energy-saving renovation projects based on BWM weighting method
  57. Deep neural network application in real-time economic dispatch and frequency control of microgrids
  58. Real-time force/position control of soft growing robots: A data-driven model predictive approach
  59. Mechanical product design and manufacturing system based on CNN and server optimization algorithm
  60. Application of finite element analysis in the formal analysis of ancient architectural plaque section
  61. Research on territorial spatial planning based on data mining and geographic information visualization
  62. Fault diagnosis of agricultural sprinkler irrigation machinery equipment based on machine vision
  63. Closure technology of large span steel truss arch bridge with temporarily fixed edge supports
  64. Intelligent accounting question-answering robot based on a large language model and knowledge graph
  65. Analysis of manufacturing and retailer blockchain decision based on resource recyclability
  66. Flexible manufacturing workshop mechanical processing and product scheduling algorithm based on MES
  67. Exploration of indoor environment perception and design model based on virtual reality technology
  68. Tennis automatic ball-picking robot based on image object detection and positioning technology
  69. A new CNN deep learning model for computer-intelligent color matching
  70. Design of AR-based general computer technology experiment demonstration platform
  71. Indoor environment monitoring method based on the fusion of audio recognition and video patrol features
  72. Health condition prediction method of the computer numerical control machine tool parts by ensembling digital twins and improved LSTM networks
  73. Establishment of a green degree evaluation model for wall materials based on lifecycle
  74. Quantitative evaluation of college music teaching pronunciation based on nonlinear feature extraction
  75. Multi-index nonlinear robust virtual synchronous generator control method for microgrid inverters
  76. Manufacturing engineering production line scheduling management technology integrating availability constraints and heuristic rules
  77. Analysis of digital intelligent financial audit system based on improved BiLSTM neural network
  78. Attention community discovery model applied to complex network information analysis
  79. A neural collaborative filtering recommendation algorithm based on attention mechanism and contrastive learning
  80. Rehabilitation training method for motor dysfunction based on video stream matching
  81. Research on façade design for cold-region buildings based on artificial neural networks and parametric modeling techniques
  82. Intelligent implementation of muscle strain identification algorithm in Mi health exercise induced waist muscle strain
  83. Optimization design of urban rainwater and flood drainage system based on SWMM
  84. Improved GA for construction progress and cost management in construction projects
  85. Evaluation and prediction of SVM parameters in engineering cost based on random forest hybrid optimization
  86. Museum intelligent warning system based on wireless data module
  87. Optimization design and research of mechatronics based on torque motor control algorithm
  88. Special Issue: Nonlinear Engineering’s significance in Materials Science
  89. Experimental research on the degradation of chemical industrial wastewater by combined hydrodynamic cavitation based on nonlinear dynamic model
  90. Study on low-cycle fatigue life of nickel-based superalloy GH4586 at various temperatures
  91. Some results of solutions to neutral stochastic functional operator-differential equations
  92. Ultrasonic cavitation did not occur in high-pressure CO2 liquid
  93. Research on the performance of a novel type of cemented filler material for coal mine opening and filling
  94. Testing of recycled fine aggregate concrete’s mechanical properties using recycled fine aggregate concrete and research on technology for highway construction
  95. A modified fuzzy TOPSIS approach for the condition assessment of existing bridges
  96. Nonlinear structural and vibration analysis of straddle monorail pantograph under random excitations
  97. Achieving high efficiency and stability in blue OLEDs: Role of wide-gap hosts and emitter interactions
  98. Construction of teaching quality evaluation model of online dance teaching course based on improved PSO-BPNN
  99. Enhanced electrical conductivity and electromagnetic shielding properties of multi-component polymer/graphite nanocomposites prepared by solid-state shear milling
  100. Optimization of thermal characteristics of buried composite phase-change energy storage walls based on nonlinear engineering methods
  101. A higher-performance big data-based movie recommendation system
  102. Nonlinear impact of minimum wage on labor employment in China
  103. Nonlinear comprehensive evaluation method based on information entropy and discrimination optimization
  104. Application of numerical calculation methods in stability analysis of pile foundation under complex foundation conditions
  105. Research on the contribution of shale gas development and utilization in Sichuan Province to carbon peak based on the PSA process
  106. Characteristics of tight oil reservoirs and their impact on seepage flow from a nonlinear engineering perspective
  107. Nonlinear deformation decomposition and mode identification of plane structures via orthogonal theory
  108. Numerical simulation of damage mechanism in rock with cracks impacted by self-excited pulsed jet based on SPH-FEM coupling method: The perspective of nonlinear engineering and materials science
  109. Cross-scale modeling and collaborative optimization of ethanol-catalyzed coupling to produce C4 olefins: Nonlinear modeling and collaborative optimization strategies
  110. Unequal width T-node stress concentration factor analysis of stiffened rectangular steel pipe concrete
  111. Special Issue: Advances in Nonlinear Dynamics and Control
  112. Development of a cognitive blood glucose–insulin control strategy design for a nonlinear diabetic patient model
  113. Big data-based optimized model of building design in the context of rural revitalization
  114. Multi-UAV assisted air-to-ground data collection for ground sensors with unknown positions
  115. Design of urban and rural elderly care public areas integrating person-environment fit theory
  116. Application of lossless signal transmission technology in piano timbre recognition
  117. Application of improved GA in optimizing rural tourism routes
  118. Architectural animation generation system based on AL-GAN algorithm
  119. Advanced sentiment analysis in online shopping: Implementing LSTM models analyzing E-commerce user sentiments
  120. Intelligent recommendation algorithm for piano tracks based on the CNN model
  121. Visualization of large-scale user association feature data based on a nonlinear dimensionality reduction method
  122. Low-carbon economic optimization of microgrid clusters based on an energy interaction operation strategy
  123. Optimization effect of video data extraction and search based on Faster-RCNN hybrid model on intelligent information systems
  124. Construction of image segmentation system combining TC and swarm intelligence algorithm
  125. Particle swarm optimization and fuzzy C-means clustering algorithm for the adhesive layer defect detection
  126. Optimization of student learning status by instructional intervention decision-making techniques incorporating reinforcement learning
  127. Fuzzy model-based stabilization control and state estimation of nonlinear systems
  128. Optimization of distribution network scheduling based on BA and photovoltaic uncertainty
  129. Tai Chi movement segmentation and recognition on the grounds of multi-sensor data fusion and the DBSCAN algorithm
  130. Special Issue: Dynamic Engineering and Control Methods for the Nonlinear Systems - Part III
  131. Generalized numerical RKM method for solving sixth-order fractional partial differential equations
https://doi.org/10.1515/nleng-2024-0082
Schlagwörter für diesen Artikel
motor function; video stream matching; rehabilitation training; action recognition
Creative Commons
BY 4.0

Artikel in diesem Heft

  1. Research Articles
  2. Generalized (ψ,φ)-contraction to investigate Volterra integral inclusions and fractal fractional PDEs in super-metric space with numerical experiments
  3. Solitons in ultrasound imaging: Exploring applications and enhancements via the Westervelt equation
  4. Stochastic improved Simpson for solving nonlinear fractional-order systems using product integration rules
  5. Exploring dynamical features like bifurcation assessment, sensitivity visualization, and solitary wave solutions of the integrable Akbota equation
  6. Research on surface defect detection method and optimization of paper-plastic composite bag based on improved combined segmentation algorithm
  7. Impact the sulphur content in Iraqi crude oil on the mechanical properties and corrosion behaviour of carbon steel in various types of API 5L pipelines and ASTM 106 grade B
  8. Unravelling quiescent optical solitons: An exploration of the complex Ginzburg–Landau equation with nonlinear chromatic dispersion and self-phase modulation
  9. Perturbation-iteration approach for fractional-order logistic differential equations
  10. Variational formulations for the Euler and Navier–Stokes systems in fluid mechanics and related models
  11. Rotor response to unbalanced load and system performance considering variable bearing profile
  12. DeepFowl: Disease prediction from chicken excreta images using deep learning
  13. Channel flow of Ellis fluid due to cilia motion
  14. A case study of fractional-order varicella virus model to nonlinear dynamics strategy for control and prevalence
  15. Multi-point estimation weldment recognition and estimation of pose with data-driven robotics design
  16. Analysis of Hall current and nonuniform heating effects on magneto-convection between vertically aligned plates under the influence of electric and magnetic fields
  17. A comparative study on residual power series method and differential transform method through the time-fractional telegraph equation
  18. Insights from the nonlinear Schrödinger–Hirota equation with chromatic dispersion: Dynamics in fiber–optic communication
  19. Mathematical analysis of Jeffrey ferrofluid on stretching surface with the Darcy–Forchheimer model
  20. Exploring the interaction between lump, stripe and double-stripe, and periodic wave solutions of the Konopelchenko–Dubrovsky–Kaup–Kupershmidt system
  21. Computational investigation of tuberculosis and HIV/AIDS co-infection in fuzzy environment
  22. Signature verification by geometry and image processing
  23. Theoretical and numerical approach for quantifying sensitivity to system parameters of nonlinear systems
  24. Chaotic behaviors, stability, and solitary wave propagations of M-fractional LWE equation in magneto-electro-elastic circular rod
  25. Dynamic analysis and optimization of syphilis spread: Simulations, integrating treatment and public health interventions
  26. Visco-thermoelastic rectangular plate under uniform loading: A study of deflection
  27. Threshold dynamics and optimal control of an epidemiological smoking model
  28. Numerical computational model for an unsteady hybrid nanofluid flow in a porous medium past an MHD rotating sheet
  29. Regression prediction model of fabric brightness based on light and shadow reconstruction of layered images
  30. Dynamics and prevention of gemini virus infection in red chili crops studied with generalized fractional operator: Analysis and modeling
  31. Qualitative analysis on existence and stability of nonlinear fractional dynamic equations on time scales
  32. Fractional-order super-twisting sliding mode active disturbance rejection control for electro-hydraulic position servo systems
  33. Analytical exploration and parametric insights into optical solitons in magneto-optic waveguides: Advances in nonlinear dynamics for applied sciences
  34. Bifurcation dynamics and optical soliton structures in the nonlinear Schrödinger–Bopp–Podolsky system
  35. Review Article
  36. Haar wavelet collocation method for existence and numerical solutions of fourth-order integro-differential equations with bounded coefficients
  37. Special Issue: Nonlinear Analysis and Design of Communication Networks for IoT Applications - Part II
  38. Silicon-based all-optical wavelength converter for on-chip optical interconnection
  39. Research on a path-tracking control system of unmanned rollers based on an optimization algorithm and real-time feedback
  40. Analysis of the sports action recognition model based on the LSTM recurrent neural network
  41. Industrial robot trajectory error compensation based on enhanced transfer convolutional neural networks
  42. Research on IoT network performance prediction model of power grid warehouse based on nonlinear GA-BP neural network
  43. Interactive recommendation of social network communication between cities based on GNN and user preferences
  44. Application of improved P-BEM in time varying channel prediction in 5G high-speed mobile communication system
  45. Construction of a BIM smart building collaborative design model combining the Internet of Things
  46. Optimizing malicious website prediction: An advanced XGBoost-based machine learning model
  47. Economic operation analysis of the power grid combining communication network and distributed optimization algorithm
  48. Sports video temporal action detection technology based on an improved MSST algorithm
  49. Internet of things data security and privacy protection based on improved federated learning
  50. Enterprise power emission reduction technology based on the LSTM–SVM model
  51. Construction of multi-style face models based on artistic image generation algorithms
  52. Research and application of interactive digital twin monitoring system for photovoltaic power station based on global perception
  53. Special Issue: Decision and Control in Nonlinear Systems - Part II
  54. Animation video frame prediction based on ConvGRU fine-grained synthesis flow
  55. Application of GGNN inference propagation model for martial art intensity evaluation
  56. Benefit evaluation of building energy-saving renovation projects based on BWM weighting method
  57. Deep neural network application in real-time economic dispatch and frequency control of microgrids
  58. Real-time force/position control of soft growing robots: A data-driven model predictive approach
  59. Mechanical product design and manufacturing system based on CNN and server optimization algorithm
  60. Application of finite element analysis in the formal analysis of ancient architectural plaque section
  61. Research on territorial spatial planning based on data mining and geographic information visualization
  62. Fault diagnosis of agricultural sprinkler irrigation machinery equipment based on machine vision
  63. Closure technology of large span steel truss arch bridge with temporarily fixed edge supports
  64. Intelligent accounting question-answering robot based on a large language model and knowledge graph
  65. Analysis of manufacturing and retailer blockchain decision based on resource recyclability
  66. Flexible manufacturing workshop mechanical processing and product scheduling algorithm based on MES
  67. Exploration of indoor environment perception and design model based on virtual reality technology
  68. Tennis automatic ball-picking robot based on image object detection and positioning technology
  69. A new CNN deep learning model for computer-intelligent color matching
  70. Design of AR-based general computer technology experiment demonstration platform
  71. Indoor environment monitoring method based on the fusion of audio recognition and video patrol features
  72. Health condition prediction method of the computer numerical control machine tool parts by ensembling digital twins and improved LSTM networks
  73. Establishment of a green degree evaluation model for wall materials based on lifecycle
  74. Quantitative evaluation of college music teaching pronunciation based on nonlinear feature extraction
  75. Multi-index nonlinear robust virtual synchronous generator control method for microgrid inverters
  76. Manufacturing engineering production line scheduling management technology integrating availability constraints and heuristic rules
  77. Analysis of digital intelligent financial audit system based on improved BiLSTM neural network
  78. Attention community discovery model applied to complex network information analysis
  79. A neural collaborative filtering recommendation algorithm based on attention mechanism and contrastive learning
  80. Rehabilitation training method for motor dysfunction based on video stream matching
  81. Research on façade design for cold-region buildings based on artificial neural networks and parametric modeling techniques
  82. Intelligent implementation of muscle strain identification algorithm in Mi health exercise induced waist muscle strain
  83. Optimization design of urban rainwater and flood drainage system based on SWMM
  84. Improved GA for construction progress and cost management in construction projects
  85. Evaluation and prediction of SVM parameters in engineering cost based on random forest hybrid optimization
  86. Museum intelligent warning system based on wireless data module
  87. Optimization design and research of mechatronics based on torque motor control algorithm
  88. Special Issue: Nonlinear Engineering’s significance in Materials Science
  89. Experimental research on the degradation of chemical industrial wastewater by combined hydrodynamic cavitation based on nonlinear dynamic model
  90. Study on low-cycle fatigue life of nickel-based superalloy GH4586 at various temperatures
  91. Some results of solutions to neutral stochastic functional operator-differential equations
  92. Ultrasonic cavitation did not occur in high-pressure CO2 liquid
  93. Research on the performance of a novel type of cemented filler material for coal mine opening and filling
  94. Testing of recycled fine aggregate concrete’s mechanical properties using recycled fine aggregate concrete and research on technology for highway construction
  95. A modified fuzzy TOPSIS approach for the condition assessment of existing bridges
  96. Nonlinear structural and vibration analysis of straddle monorail pantograph under random excitations
  97. Achieving high efficiency and stability in blue OLEDs: Role of wide-gap hosts and emitter interactions
  98. Construction of teaching quality evaluation model of online dance teaching course based on improved PSO-BPNN
  99. Enhanced electrical conductivity and electromagnetic shielding properties of multi-component polymer/graphite nanocomposites prepared by solid-state shear milling
  100. Optimization of thermal characteristics of buried composite phase-change energy storage walls based on nonlinear engineering methods
  101. A higher-performance big data-based movie recommendation system
  102. Nonlinear impact of minimum wage on labor employment in China
  103. Nonlinear comprehensive evaluation method based on information entropy and discrimination optimization
  104. Application of numerical calculation methods in stability analysis of pile foundation under complex foundation conditions
  105. Research on the contribution of shale gas development and utilization in Sichuan Province to carbon peak based on the PSA process
  106. Characteristics of tight oil reservoirs and their impact on seepage flow from a nonlinear engineering perspective
  107. Nonlinear deformation decomposition and mode identification of plane structures via orthogonal theory
  108. Numerical simulation of damage mechanism in rock with cracks impacted by self-excited pulsed jet based on SPH-FEM coupling method: The perspective of nonlinear engineering and materials science
  109. Cross-scale modeling and collaborative optimization of ethanol-catalyzed coupling to produce C4 olefins: Nonlinear modeling and collaborative optimization strategies
  110. Unequal width T-node stress concentration factor analysis of stiffened rectangular steel pipe concrete
  111. Special Issue: Advances in Nonlinear Dynamics and Control
  112. Development of a cognitive blood glucose–insulin control strategy design for a nonlinear diabetic patient model
  113. Big data-based optimized model of building design in the context of rural revitalization
  114. Multi-UAV assisted air-to-ground data collection for ground sensors with unknown positions
  115. Design of urban and rural elderly care public areas integrating person-environment fit theory
  116. Application of lossless signal transmission technology in piano timbre recognition
  117. Application of improved GA in optimizing rural tourism routes
  118. Architectural animation generation system based on AL-GAN algorithm
  119. Advanced sentiment analysis in online shopping: Implementing LSTM models analyzing E-commerce user sentiments
  120. Intelligent recommendation algorithm for piano tracks based on the CNN model
  121. Visualization of large-scale user association feature data based on a nonlinear dimensionality reduction method
  122. Low-carbon economic optimization of microgrid clusters based on an energy interaction operation strategy
  123. Optimization effect of video data extraction and search based on Faster-RCNN hybrid model on intelligent information systems
  124. Construction of image segmentation system combining TC and swarm intelligence algorithm
  125. Particle swarm optimization and fuzzy C-means clustering algorithm for the adhesive layer defect detection
  126. Optimization of student learning status by instructional intervention decision-making techniques incorporating reinforcement learning
  127. Fuzzy model-based stabilization control and state estimation of nonlinear systems
  128. Optimization of distribution network scheduling based on BA and photovoltaic uncertainty
  129. Tai Chi movement segmentation and recognition on the grounds of multi-sensor data fusion and the DBSCAN algorithm
  130. Special Issue: Dynamic Engineering and Control Methods for the Nonlinear Systems - Part III
  131. Generalized numerical RKM method for solving sixth-order fractional partial differential equations
Jetzt anmelden und 20% sparen
Institutioneller Zugang
Wie funktioniert Authentifizierung?
Haben Sie eine Idee, wie wir unsere Website verbessern können?
Bitte schreiben Sie uns.
© 2025 De Gruyter Brill
Heruntergeladen am 14.11.2025 von https://www.degruyterbrill.com/document/doi/10.1515/nleng-2024-0082/html
Button zum nach oben scrollen