Aug 2026· Journal of Mechanics in Medicine and Biology· 0 citations
TL;DR
An auxiliary action-recognition evaluation framework incorporating a Big Generative Adversarial Network (BigGAN)-based data augmentation mechanism that offers a reproducible foundation for data augmentation, action classification, and intelligent feedback in sports motion monitoring applications is developed.
Abstract
To meet the growing demand for sports motion recognition in training assistance, rehabilitation assessment, and intelligent monitoring, this study developed an auxiliary action-recognition evaluation framework incorporating a Big Generative Adversarial Network (BigGAN)-based data augmentation mechanism. An experimental subset was first constructed from representative sports actions in the Nanyang Technological University RGB+D120 (NTU RGB+D120) dataset, including running, jumping, throwing, and gymnastics movements. BigGAN was subsequently employed to generate augmented training samples, while truncation strategies and class-embedding mechanisms were introduced to increase pose variability and visual diversity. The generated samples were used exclusively during model training, whereas only real samples were retained in the validation and test sets to ensure an unbiased performance evaluation. The quality of the generated data was assessed using Fréchet Inception Distance (FID), Inception Score (IS), Learned Perceptual Image Patch Similarity (LPIPS), nearest-neighbor retrieval, and truncation-threshold sensitivity analysis. Samples exhibiting semantic inconsistencies, structural distortions, or near-duplicate characteristics were removed through quality control procedures. Experimental results showed that the proposed framework achieved an accuracy of 89.8%, a recall of 89.2%, and an F1-score of 89.5% on the test set. The generated samples yielded an average FID of 38.80, an average LPIPS value of 0.331, and a nearest-neighbor duplication rate of 3.5%. In addition, the BigGAN-enhanced model demonstrated more consistent classification performance across different sports categories. Running achieved the highest recognition accuracy, whereas throwing remained the most challenging category. The augmentation strategy improved the model's robustness to intra-class variability in several action categories. Nevertheless, the proposed framework provides only auxiliary support for action recognition and cannot replace professional assessments conducted by coaches, clinical practitioners, or biomechanical analysis systems. These findings offer a reproducible foundation for data augmentation, action classification, and intelligent feedback in sports motion monitoring applications.
An integrated multi-layer hybrid framework for accurate, real-time posture assessment in healthcare and rehabilitation contexts is proposed, although all solutions trade off accuracy, computational cost, and practical generalizability.
A. Paul, L. Damahe· African Journal Of Applied R...· 0 citations
Human Activity Recognition (HAR) using wearable sensors is relevant to rehabilitation, assistive robotics, and mobile health applications. This study presents (i) SDALLE, a publicly available multimodal dataset integrating surface electromyography (EMG) and inertial measurement unit (IMU) signals acquired using a DELSYS Trigno wireless system; (ii) a harmonized, subject-dependent comparison with the public ENABL3S dataset; and (iii) an exploratory Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP) augmentation strategy for class balancing. SDALLE contains recordings of nine healthy male subjects walking, jogging, climbing stairs and descending stairs. EMG and IMU signals were synchronized, standardized, and segmented using 75%-overlapping windows. In the implemented pipeline, feature standardization preceded segmentation, after which the windows were randomly divided into 70% training and 30% testing subsets without participant- or trial-level grouping. Consequently, overlapping windows, including windows from the same participant and trial, could occur in both subsets. Under this window-level protocol, attention-enhanced CNN–LSTM models recorded approximately 98–99% accuracy on ENABL3S and approximately 99% on SDALLE. These values are protocol-specific and may be optimistic; they do not measure generalization to unseen participants or establish a portable model ranking. GAN augmentation increased the SDALLE row count by approximately 12% and changed average testing accuracy from 96.9% before augmentation to 96.5% afterward under the implemented protocol. The study therefore provides a dataset resource and an exploratory subject-dependent comparison of predictive and computational behavior. Participant-independent evaluation with training-fold-only normalization, model selection, and GAN training is required before drawing conclusions about cross-subject robustness or deployment readiness.
M. Farouk, M. F. El-Khatib, M. Awad et al.· Scientific Reports· 0 citations
The rapid growth of artificial intelligence (AI) has shifted the development of wearable health monitoring devices toward intelligent design. However, conventional approaches still rely heavily on experience, involve inefficient user requirement acquisition, and lack systematic support for innovation. To address these limitations, this study proposes a data-driven framework for intelligent product development that integrates deep learning, Kansei Engineering, and TRIZ theory. The framework establishes an AI-assisted work-flow comprising online data acquisition, requirement identification, knowledge mapping, concept generation, and engineering optimization. First, web crawlers collect user reviews and product images. A convolutional neural network (CNN) identifies form features, while a generative adversarial network (GAN) generates diverse design concepts. Kansei Engineering then maps user requirements to design elements, and TRIZ theory resolves engineering contradictions to support systematic optimization. A case study of smart health watches, using a dataset of nearly 30,000 images and more than 12,000 reviews, demonstrates that the framework can effectively automate requirement extraction, concept generation, and design optimization. Compared with traditional experience-driven methods, the proposed framework significantly improves decision-making efficiency and supports product innovation. This study extends the application of deep learning to engineering design support and provides a transferable methodology for data-driven intelligent product development.
K. Wang, Z. Xiong, K.-X. Li et al.· Advances in Production Engin...· 0 citations
Aiming at the key problems such as the separation of action recognition and quality assessment tasks, coarse feedback granularity and high labeling cost in the automatic evaluation of rehabilitation training, this paper proposes PDDS‑Net. The framework takes human skeleton sequence as input and realizes action classification and location-level deviation location synchronously through the collaborative architecture of a global action recognition stream and a local part evaluation stream. In the local flow, the joint nodes are divided into three functional part groups: upper limb, trunk and lower limb. independent graph convolutional subnetworks are used for decoupled coding, and a contrastive learning strategy is used to drive the attention map to focus on abnormal joints without frame‑level annotations. This is mainly achieved through the adaptive fusion mechanism to dynamically integrate the confidence of the deep network and the matching score of dynamic time warping template to maintain decision stability under conditions of pose degradation. Experiments on the NTU RGB+D and PKU‑MMD public datasets show that the action recognition accuracy of PDDS‑Net reaches 93.7%, the average precision of position‑level feedback is 0.656, 82.3% of the original AUC is still maintained at a 40% joint dropout rate, and the single‑frame inference latency is 41.8 ms, which meets the requirements of real-time interaction. This study provides a replicable technical path for the validation of rehabilitation evaluation algorithms without clinical data collection.
Mingxiang Yang· Journal of Discovery Core· 0 citations
Accurate detection of motion anomalies during physical rehabilitation is important for patient safety and recovery monitoring. Traditional assessment methods rely on visual observation. This often leads to subjective and inconsistent evaluations. To address this limitation, we propose a Support Vector-Guided Decomposition (SVGD) framework for semi-supervised rehabilitation-oriented motion anomaly detection using wearable sensor data. The framework integrates low-rank and sparse matrix decomposition with margin-based discriminative learning. Multi-sensor motion signals are processed through sequential layers of preprocessing, feature extraction, decomposition, and anomaly scoring. This design enables separation of structured rehabilitation-like movements from irregular deviations and compensatory behaviors. The framework was evaluated using structured rehabilitation-like motion data with controlled perturbation-based anomaly generation under the Leave-One-Subject-Out (LOSO) validation protocol. This provides a systematic proxy evaluation in the absence of clinically annotated rehabilitation datasets. The proposed approach achieves strong performance, with a Precision of 0.97, Recall of 0.96, F1-score of 0.965, and AUC of 0.98. These results outperform both classical and deep learning baselines. The findings demonstrate that SVGD is robust, interpretable, and computationally efficient. The framework performs well even with limited labeled data. These results highlight the potential of SVGD for rehabilitation-oriented motion anomaly detection using wearable sensor data. Validation on clinically annotated rehabilitation datasets will be considered in future work.
Kumar Dorthi, Kiran Kumar Mamidi, Ravi Kanth Kotha et al.· Scientific Reports· 0 citations
Dynamic monitoring of health behaviors in adolescent sports is crucial for promoting healthy growth. Traditional monitoring methods, however, suffer from limitations such as incomplete data acquisition, low analysis efficiency, and poor sustainability. This study addresses these issues by proposing a dynamic monitoring system for adolescent sports and health behaviors. The proposed system utilizes multi-source sensor data and image extraction algorithms. A novel multimodal feature fusion algorithm based on depth motion images (DMIs) was developed for efficient sports action recognition and anomaly analysis. Depth data was converted into DMIs, and inertial data into signal images. Both image types were processed using local ternary pattern (LTP) to generate multimodal input data. Convolutional neural networks (CNNs) were used to extract features from each modality. Feature-level fusion was then performed using discriminant correlation analysis (DCA) to maximize feature correlation between modalities and minimize feature correlation within modalities and between categories. Finally, fused features were classified using a multi-class support vector machine (SVM) to achieve efficient motion behavior recognition and analysis. The experimental results demonstrate that the proposed system achieved a recognition accuracy exceeding 90% in classifying five adolescent sports behaviors. This performance surpassed that of other comparative algorithms. Furthermore, the method achieved competitive recognition accuracy across the evaluated action categories, demonstrating its applicability and robustness in complex scenarios. The findings of this study offer new insights and methodologies for monitoring and intervening in adolescent sports health behaviors. The proposed system and algorithms provide technical support for the sustainable design of dynamic monitoring systems in this field.
