Search Results (97)

To address the issues of low perception accuracy and poor robustness in traditional motion recognition methods within complex walking environments for visually impaired individuals, this study utilizes multi-modal data, including ECG, PPG, and IMU, for classification. Regarding the low filtering efficiency of multi-modal data, an improved wavelet filtering algorithm based on LSTM is proposed. To further enhance classification accuracy, this paper introduces a motion recognition method for the blindfolded mobility simulation based on an Attention-based Two-Stream Deep Fusion Convolutional Neural Network (ATS-DFCNN). The proposed method constructs a two-stream heterogeneous feature extraction architecture by synchronously collecting tri-axial motion signals and physiological signals from subjects. A 1D-CNN is employed to capture the spatial geometric features of limb movements, while a hybrid CNN-GRU network is utilized to mine the temporal evolution patterns of physiological stress. Furthermore, an attention mechanism is introduced to achieve dynamic weighted fusion at the feature level, which strengthens critical motion features and suppresses environmental noise. Experiments were conducted with 10 subjects simulating the movements of visually impaired individuals, covering typical actions such as walking, standing, climbing stairs, descending stairs, and falling. The results demonstrate that the proposed adaptive filtering algorithm achieves an AUC of 0.942, significantly improving feature distinctiveness compared to traditional algorithms. The ATS-DFCNN model achieved an average recognition accuracy of 92.2% across five activity categories, representing a 4.8% performance increase over single IMU modal classification. Particularly in fall detection, the model effectively reduces false alarms through physiological feedback and accurately infers motion intentions, providing reliable technical support for the safety monitoring of intelligent walking-aid systems. Full article

Accurate soil salinization monitoring in arid oases is crucial for agricultural sustainability and ecological security. However, existing deep learning-based approaches often suffer from insufficient use of multi-scale information and inadequate modeling of feature interactions, limiting their accuracy for retrieving complex salinity patterns. To address these limitations, we propose a multi-scale attention-based optimized hybrid deep learning model that integrates multi-scale 1D convolutional neural networks (1D-CNN), bidirectional gated recurrent units (Bi-GRU), and Transformer mechanisms (termed SMS–1D-CNN–Bi-GRU–Transformer). In this study, “scale” refers to the receptive-field scale formed by different 1D convolutional kernel sizes. The model employs a multi-scale feature extraction module to capture remote sensing signals across different scales, a multi-scale attention mechanism to adaptively weight the most informative features, and a Bi-GRU–Transformer module to explore complex sequential and global feature relationships. The proposed framework is applied to an oasis irrigation zone in Weili County, Xinjiang, using hyperspectral data from the ZY-1E satellite, topographic indices, and spectral-derived variables. The proposed method outperforms conventional 1D-CNN, GRU–Transformer, and other benchmark models on the test set—showing improvements of 2.8% in the coefficient of determination (0.952) and 18.9% in the root mean square error (0.867 g·kg⁻¹), demonstrating practical utility for precision land management and salinity monitoring in vulnerable irrigated ecosystems. Full article

Radar quantitative precipitation estimation (QPE) is a crucial product for nowcasting and disaster warning. However, its accuracy is constrained by factors such as radar band, attenuation effects, and variations in the phase and microphysical properties of precipitation particles. Based on X-band phased-array radar data from Zhongshan City, Guangdong Province, this study compares and evaluates the QPE correction performance of three deep learning models: stacking ensemble learning, gated recurrent unit (GRU), and three-dimensional convolutional neural network (3D CNN). The aim is to explore the applicability of different model types under complex precipitation conditions. Data from August 2023 to August 2024 were used to construct the samples, with records from May 2024 held out as an independent test set and excluded from model training and hyperparameter tuning. Model performance was assessed under different radar combinations (three-radar, dual-radar, and single-radar configurations), temporal scales (minute and hourly), and precipitation intensities. The results show that: (1) at the minute scale, all three models improved the original QPE, reducing average relative error (RE) by approximately 24.6–29.5%, mean absolute error (MAE) by 23.2–27.7%, and root-mean-square error (RMSE) by 19.7–22.8%, while increasing correlation coefficient (CC) by approximately 20.4–20.9%. Specifically, GRU achieved the largest reduction in RE, stacking showed slight advantages in controlling MAE and RMSE, and 3D CNN and GRU showed similar improvements in CC. (2) At the hourly scale, the correction effect varied with precipitation intensity. In the light-to-moderate rainfall range ($0.1\le R<8.0\mathrm{mm}{\mathrm{h}}^{-1}$, where R denotes hourly rainfall), 3D CNN generally showed better error-control performance, whereas the advantage of GRU was less consistent among radar combinations. In the heavy-rainfall range ($R\ge 16.0\mathrm{mm}{\mathrm{h}}^{-1}$), stacking and GRU provided complementary value in some radar configurations, although model performance remained configuration dependent. (3) Case analysis shows that stacking can improve the original QPE at some extreme-precipitation stations, but correction performance in the extreme high-value range remains unstable, and GRU and 3D CNN are more prone to underestimation. Oriented toward operational applications, this study systematically evaluates the applicability and limitations of three model types under different scenarios while considering computational-resource constraints and timeliness requirements, thereby providing a reference for model selection and operational application in radar QPE correction. Full article

Wearable inertial sensors produce multi-axis motion signals with rich spatial and temporal structure. Existing deep-learning pipelines for human activity recognition (HAR) rarely tackle three issues jointly: explicit modeling of the body-part grouping of multi-location inertial channels, bidirectional temporal modeling at linear-time cost, and dynamic, time-varying attention for non-stationary motion. We aim to close these three gaps within a single architecture. To this end we propose Dynamic-Attentive Selective Mamba (DASM), which combines three components: Group-Aware Convolutions (GroupConv) for body-part-aware local features, a Bidirectional Mamba (BiMamba) module for linear-time forward and backward temporal context, and a Dynamic CBAM (DCBAM) that produces per-timestep channel and spatial attention for non-stationary windows. On the UCI Daily and Sports Activities dataset (19 classes, 8 subjects), under stratified segment-level 5-fold cross-validation (3 seeds, 15 runs/model), DASM reaches 99.89% accuracy and F1, a 0.11% gain over CNN-BiGRU-CBAM and 0.50% over Multi-STMT; under leave-one-subject-out (LOSO), it reaches 89.34%, 1.69% above the strongest baseline. The 10.55% drop under LOSO shows that segment-level results overestimate cross-subject generalization. Ablations show small but statistically detectable gains (Cohen’s $d\in [0.4,0.7]$ per module, $d\ge 1.5$ full-vs-baseline). We therefore position the contribution as a structured architecture within a near-saturated benchmark; broader deployment claims require multi-dataset subject-independent validation. Full article

To address the challenges of scarce unmanned aerial vehicle (UAV) track samples, severe class imbalance, and high motion similarity between UAVs and birds in low-altitude radar recognition, this paper proposes a trajectory classification method integrating a TCN-Transformer model with a physics-constrained TimeGAN (PC-TimeGAN) data augmentation framework. Specifically, the PC-TimeGAN generates high-quality, kinematically compliant UAV trajectories to alleviate data scarcity and class imbalance. A multi-scale TCN-Transformer is then constructed to comprehensively extract features, utilizing multi-kernel dilated convolutions for local temporal correlations and self-attention mechanisms for global temporal dependencies, thereby improving the discrimination between UAV and bird trajectories with similar motion patterns. Furthermore, a joint loss function combining Focal Loss and Triplet Loss is employed to optimize the decision boundaries and feature space, enhancing model robustness and generalization. Experiments on a measured dataset demonstrate that, under the 15-dimensional input setting, the proposed method achieves a UAV recall of 80.00%, an FAR of 3.15%, a precision of 64.00%, and an F1-score of 0.7111. Compared to baseline methods (e.g., SVM, LSTM, GRU, Transformer, and 1D-CNN), the proposed approach significantly improves UAV recall under limited trajectory information while keeping the false-alarm rate of misclassifying birds as UAVs low. Ultimately, this method markedly enhances the comprehensive performance of rapid track-level target classification for low-altitude surveillance radars. Full article

This paper explores the feasibility of early-season crop classification based on Sentinel-2-time series using the TimeSen2Crop dataset (≈1 million pixels, 16 crops). The aim of the study was to evaluate the spectral-phenological separability of crops during the season and compare the performance of classical tabular algorithms, deep sequence models, and a seasonally oriented hybrid stacking scheme. Based on multispectral observations, a feature set was formed from 9 optical channels and 13 vegetation indices for 30 dates. F-criteria were calculated, confirming a sharp increase in interclass separability during the active vegetative growth phase and substantiating three time series truncation scenarios (early, early + mid-season, and full season). Random Forest (macro-F1: 0.46/0.74/0.75) was used as the base tabular model. LSTM, BiLSTM, GRU, 1D-CNN, and Transformer were trained in parallel, with Transformer showing the best results among the deep architectures (0.42/0.68/0.78). The main contribution of the work is a hybrid multi-layer stacking scheme combining heterogeneous base algorithms and OOF meta-features, which provides the highest quality (0.51/0.83/0.86) in all scenarios. The obtained results confirm the effectiveness of phenology-oriented selection of time windows, informative indices, and hybrid ensemble learning for improving the accuracy of early-season crop monitoring. Full article

Short-term load forecasting plays a pivotal role in modern power system operations yet it remains challenging due to the complex spatiotemporal dependencies in load data. This paper proposes a dual-head attention residual network (DARNet) that significantly advances STLF through three key innovations: (1) a hybrid encoder combining 1D-CNN and GRU architectures to simultaneously capture the local load patterns and long-term temporal dependencies, achieving a 28% better locality awareness than that of conventional approaches; (2) a novel dual-head attention mechanism that dynamically models both the inter-temporal relationships and cross-variable dependencies, reducing the feature engineering requirements; and (3) an autocorrelation-adjusted recursive forecasting framework that cuts the multi-step prediction error accumulation by 33% compared to that with standard seq2seq models. Extensive experiments on real-world datasets from three Chinese cities demonstrate DARNet’s superior performance, outperforming six state-of-the-art benchmarks by 21–35% across all of the evaluation metrics (MAPE, SMAPE, MAE, and RRSE) while maintaining robust generalization across different geographical regions and prediction horizons. Full article

Accurate stress measurement is essential to evaluating structural integrity and plays a pivotal role in the health monitoring and predicting the service life of steel infrastructures. This study proposes a deep learning approach for stress prediction based on longitudinal critically refracted (LCR) ultrasonic waves. The model integrates gated recurrent units (GRU), attention mechanisms, and one-dimensional convolutional neural networks (1D-CNN), enabling direct stress prediction from raw ultrasonic signals without the need for manual feature extraction or explicit physical modeling. To validate the approach, LCR signals were acquired using a custom-built piezoelectric ultrasonic system from 20# steel specimens subjected to uniaxial stresses ranging from 0 to 200 MPa. A dataset comprising 4200 samples was augmented to enhance training efficiency. The proposed model achieved a mean absolute error of 1.94 MPa. Generalization tests demonstrated high accuracy across diverse stress levels, with average errors below 3 MPa, highlighting the model’s robustness. This research presents an accurate, intelligent, and calibration-free ultrasonic method for stress evaluation, providing practical support for stress evaluation in steel structures under actual operating conditions. Full article

Early and accurate detection of skin cancer is critical for reducing mortality rates, particularly for malignant melanoma. Automated analysis of dermoscopic images has gained significant attention due to its potential to support clinical diagnosis and overcome the limitations of manual inspection. Motivated by challenges such as image noise, low contrast, lesion variability, and redundant feature representation, this study proposes an optimal deep hybrid framework for skin lesion detection and classification. The objective of this work is to design a robust and efficient system that integrates advanced preprocessing, precise segmentation, optimal feature selection, and accurate classification. Initially, contrast enhancement using Contrast Limited Adaptive Histogram Equalization (CLAHE) and noise reduction using Wiener filtering are applied to improve image quality. Lesion regions are then segmented using a Selective Kernel U-Net (SK-UNet), which adaptively captures multi-scale spatial information. Subsequently, discriminative color, texture, and shape features are extracted and optimized using the Fossa Optimization Algorithm (FOA) to eliminate redundancy. A hybrid one-dimensional Convolutional Neural Network–Gated Recurrent Unit (1D-CNN–GRU) classifier is employed for final classification, learning both spatial and sequential feature patterns. Experimental evaluation on the ISIC and DermMNIST datasets demonstrates that the proposed framework achieves classification accuracies of 97.6% and 95.6%, respectively, outperforming several existing methods. The results confirm that the proposed hybrid framework provides reliable, accurate, and scalable skin cancer diagnosis, highlighting its potential for assisting clinical decision-making and early detection. Full article

Network Intrusion Detection Systems (NIDS) play a crucial role in protecting network environments against cyberattacks. However, traditional NIDS rely heavily on predefined attack signatures, which limits their ability to detect zero-day attacks. Although machine learning-based intrusion detection techniques have been widely adopted in Network Intrusion Prevention Systems (NIPS), publicly available network traffic datasets often suffer from severe class imbalance, leading to biased learning and degraded detection performance. To address this issue, this study proposes data augmentation framework based on a 3D-GAN (Three-Discriminator Generative Adversarial Network). The proposed architecture integrates an autoencoder, a CNN (Convolutional Neural Network), and an LSTM (Long Short-Term Memory) network as parallel discriminators to capture the statistical, spatial, and temporal characteristics of network traffic. By jointly optimizing multiple discriminator losses, the framework enhances training stability and generates high-quality synthetic samples. Experiments were conducted on the CIC-UNSW-NB15 dataset using Random Forest-, XGBoost (eXtreme Gradient Boosting)-, and BiGRU (Bidirectional Gated Recurrent Unit)-based classifiers. Two augmented datasets were constructed to address class imbalance, containing approximately 100,000 and 350,000 samples, respectively. Among them, Dataset 2, augmented using the proposed 3D-GAN, demonstrated the most significant performance improvement. Compared to the original imbalanced dataset, the XGBoost classifier trained on Dataset 2 achieved approximately a 4% increase in both accuracy and F1-score, while reducing the false positive rate and false negative rate by approximately 3.5%. Furthermore, the optimal configuration attained an F1-score of 0.9816, indicating superior capability in modeling complex network traffic patterns. Overall, this study highlights the potential of GAN-based data augmentation for alleviating class imbalance and improving the robustness and generalization of intrusion detection systems. Full article

Human activity recognition (HAR) is essential in many applications, such as smart homes, assisted living, healthcare monitoring, rehabilitation, physiotherapy, and geriatric care. Conventional methods of HAR use wearable sensors, e.g., acceleration sensors and gyroscopes. However, they are limited by issues such as sensitivity to position, user inconvenience, and potential health risks with long-term use. Optical camera systems that are vision-based provide an alternative that is not intrusive; however, they are susceptible to variations in lighting, intrusions, and privacy issues. The paper uses an optical method of recognizing human domestic activities based on pose estimation and deep learning ensemble models. The skeletal keypoint features proposed in the current methodology are extracted from video data using PoseNet to generate a privacy-preserving representation that captures key motion dynamics without being sensitive to changes in appearance. A total of 30 subjects (15 male and 15 female) were sampled across 2734 activity samples, including nine daily domestic activities. There were six deep learning architectures, namely, the Transformer (Transformer), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Multilayer Perceptron (MLP), One-Dimensional Convolutional Neural Network (1D CNN), and a hybrid Convolutional Neural Network–Long Short-Term Memory (CNN–LSTM) architecture. The results on the hold-out test set show that the CNN–LSTM architecture achieves an accuracy of 98.78% within our experimental setting. Leave-One-Subject-Out cross-validation further confirms robust generalization across unseen individuals, with CNN–LSTM achieving a mean accuracy of 97.21% ± 1.84% across 30 subjects. The results demonstrate that vision-based pose estimation with deep learning is a useful, precise, and non-intrusive approach to HAR in smart healthcare and home automation systems. Full article

Urban pluvial flooding in highly developed basins is challenging to forecast in real time because detailed 1D–2D hydraulic models are computationally expensive, while purely data-driven approaches often lack physical consistency. This study aims to enable operational urban flood nowcasting by proposing a model-informed AI framework for short-term stream-stage and urban inundation prediction in the Bisan-dong district of Anyang, South Korea, where the Anyang and Hagui Streams frequently overflow. A gated recurrent unit (GRU) network was trained on 10 min rainfall and stream-stage observations from 2011 to 2018 and independently validated on 2019–2022 data at four gauges to forecast stream stage at lead times of 10–60 min. In parallel, an ANN–CNN inundation surrogate was trained on 864 XP-SWMM 1D–2D simulation scenarios, forced by design storms and downstream water-level boundary conditions, to produce 256 × 256 maps of maximum inundation depth. The GRU model achieved ${\mathrm{R}}^2$ and Nash–Sutcliffe efficiency values generally above 0.95, with a mean absolute percentage error (MAPE) below approximately 5% for 10–30-min lead times; performance decreased but remained useful at 60 min. The inundation surrogate reproduced XP-SWMM results with an MAPE of 8.89% for inundation area and 19.49% for grid-based depth. Together, the ANN–CNN system enables rapid generation of high-resolution flood maps and provides a practical basis for AI-assisted urban flood nowcasting and risk management. Full article

This study presents Care-MOVE, a smartphone-based application designed for continuous, passive, and unobtrusive monitoring of mobility, environmental exposure, and cognitive status in older adults within a telemedicine framework. The system integrates movement-related data collected through smartphone sensors (GPS, activity recognition, and caloric expenditure estimation) with contextual air quality information and standardized neuropsychological assessments, resulting in a comprehensive multimodal dataset (Care-MOVE Dataset). An exploratory proof-of-concept study was conducted on a subsample of 53 participants aged over 65, each monitored continuously for five days, contributing on average more than 30,000 longitudinal records. To investigate whether daily motor behavior can serve as a digital biomarker of cognitive functioning, several Machine Learning and Deep Learning models were evaluated using a Leave-One-User-Out (LOUO) cross-validation strategy. The comparative analysis included traditional classifiers (Logistic Regression, Random Forest, Gradient Boosting, K-Nearest Neighbors, and Support Vector Machines) as well as temporal deep learning architectures (1D CNN, LSTM, GRU, and Transformer). Among all of the evaluated approaches, the Support Vector Machine with RBF kernel achieved the best performance, reaching an accuracy of 98.1%, a balanced accuracy of 0.988, and an F1-score of 0.981, demonstrating robust generalization across unseen subjects. For this reason, the study was designed and presented as an exploratory proof-of-concept rather than a definitive clinical validation. This integrated approach not only enables the collection of detailed and contextualized data but also opens new perspectives for proactive digital healthcare, focused on risk prevention, improving quality of life, and promoting autonomy in elderly patients. Full article

The output of MEMS gyroscopes is highly vulnerable to ambient temperature variations, which induce temperature drift errors and degrade navigation precision. Consequently, temperature compensation for MEMS gyroscope outputs is of critical importance. To address this issue, this study proposes a novel temperature compensation model for the dual-mass vibration MEMS gyroscope (DMVMG), which integrates the TTAO-VMD, 1D-CNN-Bi-GRU-Attention, and SHAKF algorithms. The implementation process of the proposed model is as follows: firstly, the structural configuration and fundamental operating principle of the DMVMG are elaborated. Secondly, the temperature error compensation model is constructed based on the fusion of the TTAO-VMD, 1D-CNN-Bi-GRU-Attention, and SHAKF algorithms. Thirdly, the raw output signal of the DMVMG is preprocessed using the TTAO-VMD algorithm, which decomposes the signal into four distinct components, namely high-frequency noise, white noise, mixed noise, and temperature-induced noise. Subsequently, the high-frequency and white noise components are eliminated, while the mixed noise component is filtered via the SHAKF algorithm. On this basis, the 1D-CNN-Bi-GRU-Attention algorithm is adopted to establish the temperature error compensation model, with the temperature, temperature change rate, time, and temperature-induced noise as input variables. Finally, the optimized signal components are reconstructed to yield the temperature-compensated output of the DMVMG. The experimental results based on the Allan variance method demonstrate that the angle random walk (N) is reduced from 18.56 $°/\sqrt{h}$ to 0.17 $°/\sqrt{h}$, and the bias instability (B) is decreased from 32.76 °/h to 0.82 °/h, verifying the effectiveness of the proposed method. Full article

Artificial intelligence (AI) shows great promise for transforming the agriculture sector and can enable the development of many modern farming practices over conventional methods. Nowadays, AI agents and agentic AI have attained popularity due to their autonomous structure and working mechanism. This research work proposes an agentic AI framework that integrates multiple agents developed for farming land to promote climate-smart agriculture and support United Nations (UN) sustainable development goals (SDGs). The developed structure has four agents: Agent A for monitoring soil properties, Agent B for weather sensing, Agent C for disease detection vision sensing in rice crops, and Agent D, a multi-agent supervisor agent chatbot connected with the other agents. The overall objective was to connect all agents on a single platform to obtain sensor data and perform a predictive analysis. This will help farmers and landowners obtain information about weather conditions, soil properties, and vision-based disease detection so that appropriate measures can be taken on agricultural land for rice crops. For soil properties (nitrogen, phosphorus, and potassium) from Agent A and climate data (temperature and humidity) from Agent B, we deployed the long short-term memory (LSTM), gated recurrent unit (GRU), and one-dimensional convolutional neural network (1D-CNN) predictive models, which achieved an accuracy of 93.4%, 94%, and 96% for Agent A; a 0.27 mean absolute error (MAE) for temperature; and a 2.9 MAE for humidity on the Agent B data. For Agent C, we used vision transformer (ViT), MobileViT, and RiceNet (with a diffusion model layer as a feature extractor) models to detect disease. The models achieved accuracies of 95%, 98.5%, and 85.4% during training respectively. Overall, the proposed framework demonstrates how agentic AI can be used to transform conventional farming practices into a digital process, thereby supporting smart agriculture. Full article