Search Results (514)

Accurate monitoring of substrate moisture content and electrical conductivity (EC) is essential for irrigation and nutrient management in soilless strawberry cultivation. However, conventional sensor-based approaches are limited in spatial coverage. This study developed a multimodal prediction framework integrating canopy multispectral imaging (713–920 nm) with greenhouse environmental and irrigation time-series data to estimate substrate state non-destructively. Four spectral preprocessing schemes were evaluated, and the first derivative of digital number values was adopted as the primary preprocessing condition. Five regression models, including a proposed spectrum-query cross-attention long short-term memory network (SpecAtten-LSTM), were compared across six spectral input configurations and five non-spectral baselines. Substrate EC was predicted accurately from either modality alone. Extreme gradient boosting reached R² = 0.9710, and the environment-only baselines achieved comparable performance, suggesting that either modality contained sufficient information for EC prediction. For substrate moisture content, the highest performance was obtained when spectral and environmental information were combined. SpecAtten-LSTM achieved the highest accuracy (R² = 0.7463) under the full multimodal configuration, and an ablation analysis confirmed that its cross-attention and fusion modules drove this gain. Permutation importance identified relative humidity as the dominant environmental variable for both targets. The results indicate that canopy-level observations can be used to estimate root-zone substrate conditions and that spectral information provides additional value primarily for substrate moisture content prediction. Full article

Visible-infrared person re-identification (VI-ReID) aims to retrieve pedestrian images of the same identity across visible and infrared modalities, but remains challenging due to the large modality gap and unstable local correspondence. Existing methods mainly rely on visual cues, which may be insufficient when infrared images lack color and fine-grained texture information. To address this issue, this paper proposes a CLIP-Guided Progressive Body-Part Semantic Alignment Network, termed PBSA-Net. The proposed method introduces CLIP-derived textual semantics as modality-agnostic guidance for both global representation learning and local body-part feature extraction. Specifically, a global semantic branch first learns identity-level textual anchors to regularize global visual features. Then, a body-part semantic branch exploits identity-aware body-part prompt learning, multi-level feature fusion, and text-guided cross-attention to guide fine-grained local representation learning. A progressive three-stage optimization strategy is further adopted to decouple global semantic learning, body-part semantic correspondence learning, and retrieval-oriented feature optimization. Experiments on SYSU-MM01, RegDB, and LLCM demonstrate the effectiveness of PBSA-Net. It achieves 76.5% Rank-1 and 74.2% mAP on SYSU-MM01, 82.5% Rank-1 and 76.0% mAP on RegDB, and 61.8% Rank-1 and 65.8% mAP on LLCM. Ablation studies further show that the proposed body-part semantic alignment and progressive optimization provide complementary improvements. Full article

The accumulation of training and competition loads represents a critical determinant of musculoskeletal injury risk in professional team sports, yet contemporary monitoring systems remain limited by their reliance on single-domain temporal analysis that overlooks the multi-scale rhythmic patterns inherent in athletic workload signals. This study introduces FDTM (frequency-aware dual-stream temporal model), a deep learning framework that jointly encodes time-domain dependencies and frequency-domain spectral signatures from digital athlete monitoring streams to predict individual injury risk over a forward-looking seven-game horizon. The framework integrates a stacked bidirectional long short-term memory branch augmented with temporal self-attention pooling, a spectral encoding branch employing discrete Fourier transform decomposition across high-frequency (weekly), mid-frequency (bi-weekly), and low-frequency (seasonal) bands, and a cross-modal gated attention fusion module that adaptively balances temporal and spectral representations conditioned on player context. We evaluate FDTM on three heterogeneous public sports datasets spanning basketball (NBA game-log corpus 2013–2023), Australian rules football (AFL Player Workload Dataset), and soccer (SoccerMon open monitoring corpus), comprising 612 athletes and 247,830 player-game observations across ten competitive seasons. FDTM achieves AUC-ROC values of 0.858, 0.833, and 0.821 on the three datasets respectively, outperforming the strongest deep-learning baseline (FEDformer) by 2.0 to 3.3 percentage points and the strongest non-spectral baseline (TCN) by 3.2 to 4.5 percentage points while maintaining a Brier score below 0.04. Ablation studies confirm that the spectral branch contributes 5.1 percent to overall discriminative performance. SHAP attribution analyses identify high-frequency weekly components as the dominant injury-relevant signal, followed by low-frequency seasonal trends and the cumulative acute-to-chronic workload temporal feature, with gating-weight visualizations revealing dynamic modality contributions consistent with established sports science theory. Direct spectral analysis of the raw workload signal confirms that injury-preceding windows exhibit significantly elevated weekly-band power across all three datasets (Mann–Whitney U test, p < 1 × 10⁻⁷), and the architectural advantage is shown to be robust across 30 independent training seeds. These findings suggest that frequency-aware modeling may serve as a transferable methodology for sports engineering applications in injury prevention, return-to-play planning, and individualized rehabilitation, pending further external validation in female athletes and additional team sports. Full article

Healthcare and urban infrastructure are increasingly supported by Internet of Things-based sensing systems, in which heterogeneous physiological, environmental, and transmission-level data require reliable, low-latency processing. Existing works typically treat medical IoT sensing, smart-city anomaly detection, or edge-cloud offloading as isolated problems, thereby failing to support integrated sensing scenarios in shared smart environments. This paper introduces a Hybrid Edge–Cloud Intelligence Framework (HECIF) for reliable sensing and data fusion in smart healthcare and urban IoT environments. HECIF introduces modality-specific feature extraction, adaptive offloading to the edge cloud, an attention mechanism for multimodal fusion, and a reliability-weighted decision layer that incorporates sensor quality and transmission delay. The framework was tested on three publicly available datasets: the Multi-Sensor Medical IoT dataset for physiological signal classification, the UrbanIoT Anomaly dataset for urban anomaly detection, and the IoT Sensor Cloud Data Transmission dataset for offloading decision modeling, all from Kaggle. It achieved a 92.1% accuracy, 91.3% F1-score, 93.8% AUC, and 0.821 Matthews correlation coefficient in a simulated edge cloud environment, outperforming the baselines (logistic regression, random forest, XGBoost, MLP, CNN/LSTM). The framework also reduced the mean inference time to 29 ms, down from 142 ms in the cloud-only configuration, while achieving a throughput of 1150 samples per second. The results show that reliability-aware edge cloud fusion is feasible for cross-domain IoT sensing with a simulated edge cloud. However, physical device validation and real-world IoT network validation are still required before practical deployment. Full article

Low-slow-small (LSS) target recognition using multi-dimensional radar remains challenging due to weak signatures, similar kinematics, and overlapping short-term Doppler patterns. Digital-array radar provides continuous, complementary Doppler-spectrum and kinematic measurements; however, their heterogeneity in dimension, distribution, and physical meaning often makes direct fusion under-exploit discriminative complementarity and inadequately model temporal track evolution. To address this, we propose a Doppler-Kinematic Spatio-Temporal Graph Learning framework named Dual-Stream Spatio-Temporal Cross-Attention Graph Convolutional Network (DS-STCAGCN) for LSS target recognition using multi-dimensional radar observations. The method separately encodes Doppler-spectrum and kinematic features to preserve their modality-specific characteristics, fuses them through bidirectional cross-attention, captures long-range temporal dependencies via self-attention, and aggregates local frame-to-frame correlations through graph convolution on a time-ordered observation graph. On the public L-band digital-array dataset LSS-DAUR-1.0, DS-STCAGCN achieves 99.73% mean accuracy and maintains 98.64% at 5 dB signal-to-noise ratio (SNR). On the passive-radar dataset LSS-PR-1.0, it reaches 99.86% mean accuracy, demonstrating strong cross-modal generalization. This work provides an effective spatio-temporal modelling framework for multi-dimensional radar sensing and robust LSS target recognition. Full article

Underwater laser range-gated imaging (ULRGI) effectively suppresses water backscattering via time-slicing mechanisms, making it a primary modality for underwater vision. However, factors such as the inherent optical properties of water, intra-slice residual scattering, gating timing errors, and sensor noise make it difficult to separate target signals from the background. Consequently, the resulting images are generally affected by texture degradation and low contrast, severely limiting the accuracy of downstream tasks like object detection and environmental perception. To this end, we propose the use of a Wavelet-guided Multi-stage Gated Enhancement Network (WMGNet). Operating progressively across three stages, WMGNet’s first two stages employ an encoder–decoder architecture that leverages multi-scale frequency decomposition in the wavelet domain to pinpoint intra-slice scattering and decouple target signals from noise. To precisely extract fine details, we design a TextureBlock integrating feature gating (ConvGLU) and high-frequency attention (HFAttention). Additionally, a pixel-wise ground-truth guided attention module (GGAM) is introduced to optimize the precision and target-specificity of multi-stage feature fusion. Extensive comparative and ablation experiments demonstrate that the proposed WMGNet effectively eliminates scattering interference and restores texture details in underwater imaging. On our custom ULRGI dataset, it achieves state-of-the-art performance with a PSNR of 36.31 dB, an SSIM of 0.921, an MAE of 2.672, and an LPIPS of 0.060. Notably, it outperforms the second-best method by a margin of 3.06 dB in PSNR and reduces the MAE by 50.69%. Furthermore, evaluations on three public datasets confirm its robust cross-scenario generalization, yielding competitive PSNR values of 33.22 dB, 31.59 dB, and 32.06 dB, respectively. Overall, WMGNet provides a highly effective and robust solution for high-resolution underwater imaging. Full article

Multimodal 3D object detection benefits from the complementary nature of camera images and LiDAR point clouds. However, existing voxel–pixel fusion methods typically rely on relatively coarse cross-modal interactions, which limit fine-grained structural modeling and degrade performance on small safety-critical objects. To address this issue, we propose ETA-SDQF, an edge-texture-aware semantic dual-query fusion framework designed to enhance 3D perception of vehicles, cyclists, and pedestrians. The proposed method first introduces an edge-texture-aware image backbone (ETAIB) based on the discrete wavelet transform (DWT), which improves the representation of multi-scale fine-grained image features. Then, we design a dual-query-guided attention fusion (DQGAF) module, which leverages deformable attention to adaptively aggregate voxel-aligned multi-scale image features under joint semantic and edge-texture guidance. Finally, we adopt a hybrid 3D feature learning strategy inspired by PV-RCNN, combining voxel-based feature learning with PointNet-style feature abstraction for processing fused features. This design improves the utilization of voxel features enriched with image semantics, thereby facilitating more reliable 3D object proposal generation. Experimental results on the KITTI dataset demonstrate that the proposed framework achieves better performance compared to existing baseline methods. It consistently improves pedestrian and cyclist detection, while maintaining competitive performance on car detection across different difficulty levels, showing potential benefits on challenging KITTI samples. Full article

Human activity recognition has become an important component of human–computer interaction and ubiquitous computing. Among various sensing technologies, WiFi-based gesture recognition has attracted increasing attention due to its contactless nature and robustness to visual occlusion. However, environmental variations and user-specific differences often lead to significant performance degradation, particularly in cross-user scenarios. Existing methods primarily focus on learning domain-invariant representations, which may overlook user-specific characteristics that are essential for accurate recognition. To address this issue, we propose the Domain-invariant and User-Adaptive Learning Network (DUAL-Net), a dual-branch framework that jointly models domain-invariant and user-adaptive representations. Specifically, DUAL-Net incorporates a contrastive fusion learning (CFL) module with modality-specific encoders to learn complementary representations from WiFi and vision modalities. Furthermore, a spatial matrix difference (SMD)-guided cross-modal generation (CMG) module is introduced to generate user-adaptive WiFi features by incorporating structural priors derived from skeletal representations. To improve deployment efficiency, DUAL-Net adopts a two-stage learning framework, where adaptation is conducted offline to reduce online computational overhead. Experiments on the MM-Fi dataset and a self-collected dataset show that DUAL-Net achieves superior cross-user recognition performance compared with existing single-modality and multimodal methods. In addition, SMD-guided conditioning improves recognition accuracy by up to 8.79% over diffusion generation without structural guidance. Full article

Existing seizure detection methods cannot fully exploit the spatiotemporal features of multimodal signals. They also fail to capture deep associations among cross-modal features. This limits their ability to learn unified representations of spatiotemporal dependencies. This work proposes CMEpiNet (Complex-valued Multimodal Epilepsy detection Network model) to address this issue. CMEpiNet first uses complex-valued convolutions for feature extraction. It explicitly models phase synchronization, phase shifts, and cross-frequency coupling. Thus, EEG, ECG, and EMG features are represented in the complex-valued domain. During feature fusion, CMEpiNet uses a two-level semantic alignment-based fusion method. It applies cross-modal consistency constraints in a shared alignment space. It also performs distribution-level alignment in an epilepsy-related semantic latent space. These operations ensure the consistency of multimodal features in the global semantic structure. Finally, CMEpiNet uses a spatial attention-guided 3D convolutional classifier. The classifier jointly models the temporal, feature, and modality dimensions. Experimental results on the SeizeIT2 dataset show that CMEpiNet improves seizure detection sensitivity, reduces the false alarm rate, and maintains stable performance under perturbations. Full article

Audio–visual speech enhancement aims to recover clean speech by jointly using noisy acoustic signals and synchronized visual cues. Although diffusion-based methods achieve promising restoration performance, their multi-step sampling causes high inference latency and computational cost, limiting real-time deployment. To address this issue, this paper proposes FAV-DenoiseNet, a two-stage framework based on discriminative prior denoising and conditional residual flow matching. The first stage uses a pre-trained discriminative denoising network to suppress dominant noise and provide a structurally stable speech prior. The second stage reformulates enhancement as residual compensation between the first-stage output and the clean speech spectrum instead of directly predicting the entire clean spectrum. A conditional flow-matching network estimates the residual from zero-residual initialization through single-step inference, reducing generative sampling cost. Multi-scale cross-modal attention provides adaptive visual guidance for audio refinement at different resolutions. A residual-controlled fusion strategy preserves the stable structure recovered by the first stage while compensating for residual noise, high-frequency details, and weak speech components. The experimental results show that FAV-DenoiseNet achieves PESQ, ESTOI, and SI-SDR scores of 2.805, 0.775, and 12.480 dB on VoxCeleb2 and 3.157, 0.876, and 13.281 dB on GRID, respectively, with an RTF of 0.086. These results demonstrate that the proposed framework effectively balances enhancement quality, detail restoration, and real-time inference efficiency. Full article

To address cross-subject domain shift and insufficient complementary fusion of time-frequency information in EEG-based emotion recognition, this paper proposes a multi-task adversarial network: Path Adversarial Dual-Branch Network for EEG Emotion Recognition (PADB-Net). The model adopts a dual-branch parallel architecture for time and frequency domains, processing raw EEG waveforms and differential entropy features respectively, and extracts discriminative features using lightweight depthwise separable convolutions and channel attention. A path adversarial module is introduced for the first time in emotion recognition to align time-domain and frequency-domain feature distributions, solving the single-branch dominance problem in dual-branch fusion. Together with a domain adversarial module, the overall distributions of source and target domains as well as the internal distributions of the two modality branches are aligned within a unified framework. Experiments on a dataset containing healthy subjects and patients with major depressive disorder show that the full model significantly outperforms single-adversarial and non-adversarial baselines in accuracy, AUC, F1-score, sensitivity, and specificity, verifying the synergistic gain of the dual-adversarial mechanism. On the HybridBCI dataset, PADB-Net achieves 77.80% accuracy, 84.50% AUC, and 79.40% F1-score with only 6.45 K trainable parameters. When transferred to the public SEED dataset for three-class emotion recognition, the model attains F1-scores of 71.83% (negative), 68.99% (neutral), and 73.37% (positive), demonstrating strong cross-dataset generalizability. Full article

Vertically linked commodity markets—global futures, regional spot, and farm-gate prices—transmit information through directed cross-market channels whose strength varies with latent volatility regimes. Standard deep learning forecasters absorb both the directed cross-market dependence and the regime dependence of intrinsic-mode-aligned latent components into shared model weights, with no explicit architectural mechanism that exposes either as an inspectable structure. This paper proposes HVB-RA, a modular framework that combines two such mechanisms with a per-tier Variational Mode Decomposition and bidirectional LSTM backbone: (i) a directed cross-market attention layer in which the upstream-to-downstream topology is supplied from domain knowledge and the time-varying upstream-source attention intensities at the farm-gate tier (the regional-spot tier, with a single upstream key, reduces algebraically to a fixed residual upstream fusion) are extracted from data, and (ii) a regime-informed modal-weighting layer that mixes two trainable softmax weight profiles over IMF-aligned latent components through a filtered Markov-switching state probability fitted in a separate stage. An auxiliary post hoc projection enforces an exact linear constraint defined by long-run sample-mean ratios across tiers; the paper does not claim that these descriptive ratios are cointegrating relations or equilibrium coefficients. The framework is evaluated on three tiers of daily natural-rubber prices spanning 2038 trading days, against three external benchmarks (random walk, ARIMA(2,0,2), and an exogenous-only LSTM) and a contemporary neural hierarchical-interpolation forecaster (NHITS). Root mean squared error is reported per tier-horizon cell; a decision-aware income-smoothing metric quantifies the operational value of $h=5$ farm-gate forecasts under a 5-day selling rule; and a within-method comparison evaluates the marginal contribution of the auxiliary constraint projection. On the present single-regime test window, HVB-RA attains a lower point error than the contemporary NHITS baseline at every tier-horizon cell, while no method—including HVB-RA—improves on the random-walk floor at most cells; the regime-conditional components of the architecture are not identifiable because every calibration and test origin is classified as a high-volatility regime by the trained Markov-switching model. The paper contributes to machine learning and knowledge extraction by demonstrating how time-varying upstream-source attention intensities at the farm-gate tier and regime-dependent latent-component-weight profiles—two forms of latent structure typically absorbed into model weights—can be exposed as explicit, inspectable, and individually testable components of a multi-tier forecasting architecture, and by providing a reproducibility package documenting the conditions under which each component is expected to be identifiable. Full article

With the rapid development of cross-border collaboration, intelligent port construction, and international logistics networks, large volumes of multisource heterogeneous data are continuously generated during cross-border circulation. To address the limitations of traditional financial review and compliance auditing methods in characterizing multisource signal coupling, as well as the tendency of conventional deep models to rely on spurious correlated features with insufficient interpretability, a multisource sensing signal fusion and causally explainable risk identification framework is proposed for cross-border trade anomaly detection. In this framework, electronic trade texts, structured financial declaration fields, GPS/AIS trajectories, port weighing records, RFID data, electronic seal status, X-ray inspection images, cold-chain temperature and humidity records, and vibration data are uniformly modeled as multisource sensing signals in cross-border trade and circulation processes. Subsequently, collaborative representation among textual semantics, attribute fields, logistics status, device records, and entity relationships is achieved through a cross-modal alignment mechanism. On this basis, an engineering-constraint-guided causal risk representation module is designed to reduce the interference of spurious correlated factors, such as regions, ports, transportation modes, and textual styles, in model decisions. Meanwhile, a counterfactual anomaly response module is introduced to analyze the influence of key variable changes on risk outputs, thereby enhancing the model’s ability to identify and explain true anomaly-driving factors. Experimental results show that the proposed method achieves the best overall performance in the cross-border trade anomaly detection task, with Accuracy, Precision, Recall, F1-score, AUC, and PR-AUC reaching $0.927$, $0.842$, $0.811$, $0.826$, $0.958$, and $0.817$, respectively, clearly outperforming baseline models including Logistic Regression, Random Forest, XGBoost, BERT, BERT+MLP, and Multimodal Transformer. In cross-time, cross-region, cross-port, and cross-entity testing scenarios, high F1-score and AUC values are still maintained. Under complex conditions such as text noise, missing modalities, logistics trajectory perturbations, and missing sensing records, only limited performance degradation is observed. Ablation experiments further verify the effective contributions of cross-modal attention, contrastive alignment, causal financial debiasing, counterfactual response, and engineering constraints to performance improvement. Full article

Accurate prediction of protein–protein interaction (PPI) interface residues is essential for understanding molecular recognition and supporting structure-guided design. To integrate contextual sequence representations with structure-related physicochemical information, we propose a multimodal framework termed MD-Transformer. The model combines residue-level ProtBERT embeddings with physicochemical descriptors, including B-factor, solvent-accessible surface area (SASA), and hydrophobicity. A hybrid fusion module first aligns heterogeneous features, followed by Transformer encoding and cross-modal attention for multimodal integration. Using the DB5.5 benchmark, physicochemical descriptors were Z-score normalized exclusively with training-set statistics. Under the complex-level split protocol (Official A), MD-Transformer achieved an AUPRC of 0.564, outperforming the ablation model without physicochemical descriptors by 0.159 and reducing false-positive predictions on exposed non-interface residues. Under the homology-aware split protocol (Official B v1), the model maintained an AUPRC of 0.480 and an MCC of 0.242, indicating retained predictive capability under reduced sequence similarity constraints. Under the same aligned evaluation workflow, PeSTo achieved an AUPRC of 0.264. Further SASA-stratified analyses identified SASA as a major contributor to suppressing false-positive predictions across residue exposure environments, while also revealing a precision-recall trade-off in highly exposed residues. These results suggest that contextual sequence representations and residue-level physicochemical descriptors provide complementary predictive signals. Full article

Feature alignment and decision-level fusion in multimodal BCI–VR interaction were investigated using Transformer-based cross-modal embeddings, Lab Streaming Layer time synchronization, attention masks, and wavelet filtering for robust representation. A four-modal acquisition and synchronization platform covering electroencephalography, electromyography, eye-tracking, and speech was constructed, and fusion was achieved by introducing a stacking meta-learner together with a confidence-aware dynamic weighting mechanism. Prototype validation and comparative evaluations were conducted on virtual reality (VR) target-selection, trajectory-following, and object-manipulation tasks. The results showed that the proposed approach outperformed baselines such as weighted voting and independent single-modality classifiers in accuracy, cross-session and cross-subject generalization, and noise robustness, while achieving a measurable reduction in end-to-end response latency, indicating that an integrated semantic alignment–adaptive fusion pipeline enhanced stable outputs and robustness in multimodal interaction. The unified semantic alignment model tailored to BCI–VR can be used for establishing an integrated engineering workflow spanning synchronization, robust representation, and adaptive fusion, and for providing transferable evaluation metrics and application paradigms that offer methodological and technical references for scenarios such as rehabilitation training, virtual education, and intelligent control. Full article