Search Results (284)

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

Learning-based approaches have significantly advanced the capabilities of Simultaneous Localization and Mapping (SLAM) systems, particularly in challenging environments characterized by noise, dynamic objects, and perceptual ambiguity. However, the literature remains highly heterogeneous in terms of sensing modalities, datasets, evaluation protocols, and reporting practices, making systematic comparison difficult. This paper presents a taxonomy-driven review of learning-based SLAM approaches, with particular emphasis on LiDAR-based systems in mobile robotics, and introduces a functional taxonomy that categorizes methods according to the role of learning within the SLAM architecture: (i) learning-enhanced front-end SLAM (T1), (ii) learning-enhanced back-end SLAM (T2), and (iii) learning-centric SLAM systems (T3). Representative studies were analyzed with respect to performance characteristics, robustness, computational requirements, datasets, and deployment-related evidence. The analysis shows that T1 approaches primarily improve local pose estimation and robustness, T2 methods enhance global consistency through learning-based loop closure and relocalization, and T3 approaches explore unified representations, semantic reasoning, and learning-centric autonomy, albeit with greater computational demands and limited deployment evidence. The review further indicates that hybrid approaches combining geometric and learning-based components constitute a prominent trend in the literature, frequently reporting improvements in accuracy and adaptability while maintaining compatibility with established SLAM frameworks. Nevertheless, these observations should be interpreted cautiously, as stronger empirical evidence for hybrid systems may partially reflect their greater technological maturity and broader evaluation history. Finally, the review identifies persistent challenges, including limited cross-domain generalization, high computational requirements, limited deployment-oriented evaluation, and the lack of standardized benchmarking and reporting practices. These findings highlight the need for more reproducible evaluation methodologies, uncertainty-aware learning strategies, and computationally efficient architectures for robust real-world autonomous SLAM. Full article

DRAM-based in-memory computing integrates computational regions into the main memory, enabling local data processing within the memory, thereby achieving faster and more efficient data computation. However, enhancing system performance requires addressing a critical challenge: achieving more general and sufficiently powerful data processing capabilities within DRAM-PIM. Existing DRAM-PIM implementations often suffer from limited computational capabilities due to the shared standard DRAM package area between memory cells and computational circuits or because the operator circuits are overly customized, which limits their ability to meet required data processing demands. To address this issue, in this paper, we propose a Super Memory Processing Unit (SMPU). The SMPU uses Hybrid Bonding technology to 3D-stack DRAM and many-core computational clusters, enabling large-bandwidth (0.25 TB/s per-bank, 2 TB/s for 8-bank system bandwidth) on-chip data transmission between DRAM and the computational cluster via copper interconnects, effectively breaking the memory wall bottleneck of existing computing architectures. The SMPU constructs a dual-channel fine-grained computational cluster at the logical computing layer, providing flexible and ample computility for various AI models, such as ResNet50 and Llama2. The SMPU uses standard DDR protocols and integrates a new memory space allocation and parsing controller to ensure system compatibility without modifying the host-end hardware, facilitating the integration and invocation of computility in memory particles. Additionally, the SMPU features an independent dual-channel memory-management mechanism within the memory particles, enabling simultaneous multi-channel, multi-modal AI model inference. We compared a CPU system equipped with an SMPU to current computing systems using FPGA simulations. The FPGA simulation results show that, under the same computational configuration, the system with the SMPU improves the performance of ResNet50-v1.5 by up to 5.1× and Llama by up to 27.43× compared to the base system, while reducing system power consumption by 71.6% (ResNet50-v1.5) to 77.8% (Llama 7B). Full article

Background/Objectives: University students commonly experience psychological distress driven by academic demands, social transitions, and financial pressures. Mindfulness-based interventions have emerged as scalable approaches to improve mental health. However, evidence comparing their effectiveness across delivery formats remains limited. This systematic review aimed to evaluate the effectiveness of mindfulness-based interventions in reducing stress, anxiety, and depression and to compare outcomes across in-person, digital, and hybrid modalities. Methods: This review followed PRISMA 2020 guidelines and included randomized controlled trials (RCTs) published between January 2020 and December 2025 on mindfulness-based interventions among university students aged 18 years and older. Intervention duration ranged from 3 days to 12 weeks, with most lasting 4 to 8 weeks, and outcomes included validated measures of stress, anxiety, or depression. Literature research was conducted in PubMed, PsycINFO, CINAHL, Scopus, and Web of Science, and two reviewers independently screened studies, extracted data, and assessed methodological quality using the Joanna Briggs Institute checklist. Results: A total of 24 RCTs were included, with the highest representation from the United States and China (n = 4 each), followed by the United Kingdom and Canada. Mindfulness-based interventions demonstrated consistent reductions in depression and generally positive effects on anxiety, while effects on stress were more variable. Digital interventions demonstrated effectiveness comparable to in-person programs, though outcomes varied by intervention structure and level of guidance. Conclusions: Mindfulness-based interventions are effective in improving mental health among university students, particularly for depression and anxiety. Multi-week programs and guided digital delivery appear to enhance effectiveness and scalability. Full article

Accurate distance measurement in outdoor environments remains a challenging problem for mobile augmented reality (AR) systems due to sensor noise, environmental variability, and the limitations of single-modality approaches. Existing consumer AR solutions often prioritize usability over metric robustness, leading to performance degradation in large-scale or heterogeneous outdoor scenarios. This work presents EfMAR, an adaptive framework for outdoor mobile AR-based geospatial measurements that integrates multiple sensing modalities through a structured sensor fusion architecture. EfMAR combines visual SLAM, inertial sensing, depth information, and global positioning cues to improve robustness and consistency in distance estimation across diverse outdoor conditions. Beyond implementation, the framework formalizes a reusable architectural model for adaptive multi-sensor fusion, supporting reproducibility and future comparative research. A dedicated dataset is described, comprising 584 unique real-world evaluation instances collected across representative outdoor scenarios. External literature-derived data were utilized strictly as calibration baselines for modeled operational degradation profiles, maintaining methodological transparency. Performance evaluation focuses on analyzing relative behavior, stability, and variability across sensing approaches rather than establishing absolute accuracy benchmarks. Comparative results across multiple distance ranges and environments indicate that hybrid sensor fusion strategies exhibit more stable and consistent performance trends compared to single-modality solutions, particularly in challenging urban contexts. Dispersion analysis further highlights the influence of environmental factors such as lighting conditions and spatial scale on measurement variability. Overall, the results position EfMAR as a flexible and adaptive framework designed to enhance robustness in outdoor AR-based geospatial measurement tasks. By emphasizing consistency, transparency, and architectural generalization, this work contributes a practical foundation for future research and development in mobile AR sensing for real-world outdoor applications. Full article

Robust unmanned aerial vehicle (UAV) detection in real low-altitude anti-UAV scenarios remains challenging due to motion blur, extreme illumination, cluttered backgrounds, and tiny target sizes. Most existing UAV detectors rely on RGB imagery, but their performance often degrades severely under these adverse conditions. Event cameras, as a neuromorphic sensing modality, capture motion-sensitive responses with high temporal resolution and thus provide complementary cues for robust UAV detection. However, existing RGB–event fusion detectors usually employ homogeneous feature extraction and generic fusion mechanisms, which are insufficient to handle heterogeneous modality degradation and exploit reliable cross-modal cues. To address this limitation, we propose SMG-UAV, a sparse mutual guided RGB–event fusion network for robust small-UAV detection. The proposed method integrates a hybrid dual-branch backbone for modality-specific representation learning, a Sparse Mutual Guided Bridge for bidirectional sparse cross-modal refinement, and a Selective Gated Pyramid Neck for multiscale enhancement of weak UAV responses. Experiments on the Florence RGB-Event Drone Dataset (FRED) and the Neuromorphic-RGB Drone Detection Dataset (NeRDD) demonstrate that SMG-UAV achieves state-of-the-art performance, outperforming the strongest competing method by an average of 5.2 points in ${\mathrm{AP}}_{50}$, while delivering stronger robustness under multiple challenging anti-UAV conditions. Full article

The interaction between light and chiral plasmonic metasurfaces provides a powerful mechanism for controlling polarization states at the nanoscale. Utilizing displacement Talbot lithography for large-area fabrication, we characterized the chiroptical response by measuring the evolution of Stokes parameters to quantify phase retardation between orthogonal polarization components. To elucidate the underlying physical mechanism, we employ a hybrid finite element method and rigorous coupled-wave analysis approach to investigate the behavior of the far-field and local-field configurations. Our results reveal that the phase shift is highly sensitive to symmetry-breaking features, where the interplay between different modes dictates the overall circular dichroism signal. Furthermore, the analysis of local field plots suggests specific contributions of plasmonic modes to the chiroptical response. We conclude that the phase shift effects, characterized via Stokes parameters and modal analysis, provide a robust metric for engineering chiroptical properties in these systems. This work establishes a fundamental framework for developing compact polarization-control elements and enhances the understanding of phase-modulated light-matter interactions in chiral plasmonic metasurfaces. Full article

Graphene–plasmonic electro–optic (EO) modulators have attracted significant interest for compact and energy-efficient integrated photonic systems due to their electrically tunable optical response and strong light–matter interaction. In this work, an ultra-thin silicon-on-insulator (SOI) graphene–plasmonic slot modulator (G-PSM) is investigated using a combined semi-analytical and numerical framework. The analysis integrates finite-temperature Kubo conductivity modeling, perturbation-based effective-index analysis, overlap-factor evaluation, eigenmode analysis, and full-wave simulations to study the influence of silicon thickness on the EO performance of the proposed structure. The obtained results demonstrate that geometry engineering strongly affects modal confinement, overlap enhancement, effective-index perturbation, transmission characteristics, extinction ratio (ER), insertion loss (IL), energy-per-bit consumption, and EO bandwidth. Under optimized operating conditions, the proposed G-PSM achieves an effective refractive-index variation of approximately $3.1\times {10}^{-3}$, an ER of approximately 3.5 dB, an IL of 1.5–2 dB, an energy-per-bit consumption of approximately 7.5 fJ/bit, and a 3 dB EO bandwidth approaching 200 GHz. Strong electromagnetic confinement is achieved inside the plasmonic slot region near the graphene-active layer, enabling efficient electro–absorptive and electro–refractive modulation. Excellent agreement between the semi-analytical calculations and numerical simulations validates the developed framework and confirms the suitability of the proposed ultra-thin SOI G-PSM for compact broadband EO modulation in future integrated photonic systems. Full article

Accurate multi-horizon Significant Wave Height (SWH) forecasting is vital for offshore safety and efficiency. Beyond scheduling maintenance windows, reliable lead-time predictions provide critical early warnings to protect personnel and high-value assets from hazardous high-wave conditions. However, the non-stationary and multi-scale nature of sea states poses challenges for consistent long-term accuracy. To address this challenge, we propose a robust three-stage framework for decomposition, feature selection, and multi-horizon forecasting. Specifically, Optimal Variational Mode Decomposition (OVMD) is adopted to construct multiscale and multi-view representations of nonlinear SWH sequences, while a Triangulated Maximally Filtered Graph (TMFG) constructs a sparse dependency network to select informative and non-redundant predictors from decomposed components and environmental variables. A hybrid prediction model then combines a Temporal Convolutional Network (TCN) for local multi-scale patterns with a Bidirectional Gated Recurrent Unit (BiGRU) for long-range dependencies. Experiments on real-world buoy observations show that the proposed approach improves accuracy and robustness over commonly used statistical and deep-learning baselines across short-, medium-, and long-term horizons. Ablation studies confirm that integrating modal decomposition with sparse feature selection enhances model robustness, offering reliable decision support for offshore window planning and high-wave condition monitoring. Full article

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

Introduction: Three-dimensional deep learning offers promise for automated accurate brain tumor segmentation and survival prediction but requires robust validation across multiple MRI modalities to be effectively implemented in clinical practice. Methods: This study presents a comprehensive 3D deep learning framework using 369 cases from the BraTS2020 dataset. A 3D U-Net architecture was developed for tumor segmentation utilizing combined imaging data and optimized for computational efficiency and memory. The final 3D U-Net model segmentations were used to build machine learning 6-month and 12-month survival classifiers. Segmentation models were evaluated using multiple metrics, including the Dice Similarity Coefficient, Hausdorff Distance, and Cohen’s d. The classification models were evaluated using AUC-ROC and balanced accuracy. Results: Segmentation achieved a modest, but promising, performance across 30 epochs and with 295 training patients, achieving the best mean validation Dice = 0.8388 and a final-epoch mean Dice of 0.8263. Survival classification with a hybrid clinical and imaging logistic regression showed promising results, with 12-month prediction achieving AUC = 0.746 and 69% accuracy. The top contributing features for the 12-month prediction classifier were extent of resection, T1 contrast-enhanced tumor median, and FLAIR tumor median. Conclusions: This comprehensive framework demonstrates that a multi-modal approach provides meaningful performance gains, while segmentation-derived features show a promising ability to enable survival prediction. Full article

Seismic isolation systems are widely adopted in bridge engineering to reduce earthquake-induced force transfer and improve structural resilience. Conventional lead rubber bearings (LRBs) provide effective energy dissipation and period elongation; however, their limited recentering capability may result in significant residual displacement after strong ground motions. This study investigates the seismic performance of a two-level shape memory alloy–lead rubber bearing (TL-LRB-SMA) hybrid isolation system for multi-span bridges considering structural flexibility, support compliance, and geometric irregularity. A nonlinear analytical model of the hybrid isolator was developed and validated under cyclic loading using benchmark hysteretic behavior from the literature. Subsequently, a multi-degree-of-freedom numerical model of an eleven-span benchmark bridge was established and verified through modal analysis, equivalent static analysis, and comparison with MSBridge software (MSBridge Beta 1.0.1). Nonlinear time-history analyses were performed using multiple excitation scenarios, including the 1940 El-Centro record, Kobe ground motion, oblique seismic incidence, and combined loading cases. Flexible foundation conditions were represented using equivalent translational soil springs. The results indicate that the TL-LRB-SMA system consistently improves self-centering performance and significantly reduces residual displacement relative to conventional LRBs. For the regular bridge with 48 ft piers, residual displacement decreased from 0.786 inches to 0.268 inches under El-Centro excitation, while under combined excitation it reduced from 0.264 inches to 0.087 inches. For irregular bridge configurations, substantial residual displacement reductions were also observed under both longitudinal and oblique loading. Although moderate increases in peak displacement occurred in some cases due to staged SMA activation, the overall recentering performance improved markedly. Overall, the proposed TL-LRB-SMA system demonstrates strong potential for enhancing seismic resilience and post-earthquake serviceability of bridge structures, particularly in flexible and irregular configurations. Full article

Timely landslide detection and rapid qualitative assessment are fundamental to effective warning systems, hazard management, and risk mitigation. Yet, current practices that rely on on-site surveys and manual expert assessment remain risky, costly, and time-consuming. These limitations result in substantial delays between the event and the availability of actionable information. This study proposes a hybrid, multi-model framework that fuses RGB remote-sensing imagery with geospatial layers to enable timely landslide detection and actionable reporting. The pipeline couples an enhanced SegFormer (denoted as SDF-SegFormer-B2) model for landslide localization, a feature extraction technique for per-slide geo-attribute computation, and a lightweight instruction-tuned LLM (Mistral-7B-Instruct-v0.3) for structured, expert-style reporting. Although a few previous studies have explored landslide captioning, to our knowledge this is the first framework designed to generate structured technical reports enriched with terrain-context interpretation and qualitative intervention-priority indicators. Experiments use 26,758 georeferenced RGB tiles (64 × 64) with 3 m of spatial resolution from PlanetScope satellite imagery over Emilia–Romagna, Italy, with 68,592 annotated landslide boxes collected after the May 2023 rainfall events (~200 mm in 48 h on 1–3 May; 200–250 mm in 48 h on 16–17 May). The proposed SDF-SegFormer-B2 segmentation model achieved a precision of 85.54%, recall of 72.31%, and an F1-score of 78.39% on the unseen test dataset. To evaluate the quality of the generated landslide reports, 100 images were selected for domain-expert assessment. Among these, 58% of the reports were rated as “Very Good,” 30% as “Good,” 8% as “Acceptable,” and 4% as “Poor.” When considering only reports with complete and accurate inputs, 81.48% were rated “Very Good,” and 96.30% were rated either “Good” or “Very Good.” By integrating complementary models and modalities, the proposed approach automates localization-to-reporting and enables the generation of terrain-aware landslide summaries that may support preliminary decision-making and rapid post-disaster screening. Full article

The integration of heterogeneous healthcare data sources remains a major challenge in developing reliable and personalized predictive systems for digital healthcare applications. Traditional machine learning methods perform well on structured clinical data but often fail to effectively exploit multimodal information, while deep learning approaches may suffer from instability, weak generalization, and poor calibration when dealing with limited modalities. To address these limitations, this study proposes HA-PI-MADT, a hybrid adaptive healthcare-informed multimodal digital twin-inspired framework that combines deep multimodal representation learning with ensemble-based predictive modeling for robust and trustworthy healthcare prediction. The proposed framework integrates wearable sensor signals, electronic health records (EHRs), CT/MRI imaging representations, and population-level risk prototypes derived from the UCI diabetes dataset within a unified multimodal healthcare representation architecture. In addition, a modality-aware adaptive fusion mechanism dynamically adjusts the contribution of each modality according to its relevance and data quality, while a hybrid stacking strategy combines deep multimodal embeddings with classical ensemble learners to improve predictive robustness and ranking performance. To enhance clinical trustworthiness, calibration-aware optimization is incorporated to improve probabilistic reliability and uncertainty estimation. Extensive experiments conducted on a multimodal healthcare dataset demonstrate that HA-PI-MADT achieves a balanced performance profile across discrimination, ranking, and calibration-oriented evaluation metrics compared with several unimodal, multimodal, and ensemble baselines. The proposed framework achieves strong ranking-oriented and classification performance, including the highest AUPRC (0.6388) and F1-score (0.6327), while also demonstrating competitive calibration-oriented reliability through lower Brier score and negative log-likelihood values. The results demonstrate the effectiveness of the proposed hybrid adaptive multimodal digital twin-inspired framework for reliable, robust, and clinically trustworthy healthcare prediction. Full article

RGB–3D industrial anomaly detection seeks to jointly exploit texture and geometric cues for robust defect inspection. However, existing multimodal fusion methods still face two practical limitations: modality-specific anomaly evidence is often weakened after direct fusion, and image-level decisions remain unstable on difficult categories. To address these issues, this study develops a reliability-aware scoring enhancement on top of the released Hybrid Fusion/M3DM memory-bank pipeline. The method constructs a disagreement cue from RGB and point-cloud anomaly responses to enhance suspicious local regions and introduces a dual-branch image-level score calibration that combines a sensitive fusion branch with a robust statistical branch. Evaluated on MVTec 3D-AD under the official released-code full setting, the proposed method achieves 0.800 image-level ROCAUC, 0.980 pixel-level ROCAUC, and 0.926 AU-PRO, compared with 0.779, 0.975, and 0.915 for the corresponding released-code baseline in our environment. Additional evaluation on Eyecandies improves pixel-level ROCAUC and AU-PRO, while showing that image-level calibration remains dataset-sensitive. On a supplementary three-category Real-IAD D3 subset, the mean image-level ROCAUC, pixel-level ROCAUC, and AU-PRO improve from 0.963, 0.979, and 0.921 to 0.980, 0.988, and 0.941, respectively. These results indicate that explicit cross-modal disagreement modeling improves localization consistency, while image-level score calibration provides dataset-dependent gains rather than a uniform cross-dataset guarantee. Full article