Search Results (201)

Adversarial robustness in 3D point cloud recognition models is a critical concern in remote sensing applications, such as autonomous driving and infrastructure monitoring. Existing adversarial attack methods can compromise model performance; moreover, they often neglect the intrinsic geometric properties of point clouds, leading to perceptually unnatural perturbations that limit their practicality for robustness evaluation in real-world scenarios. To address this, we propose CA-Adv, a novel curvature-adaptive weighted adversarial generation method for 3D point clouds. Our approach first employs Shapley values to assess regional sensitivity and identify salient regions. It then adaptively partitions these regions based on local curvature and assigns perturbation weights accordingly, concentrating the attack on geometrically sensitive areas while preserving overall structural consistency through explicit geometric constraints. Extensive experiments on real-world remote sensing data (KITTI) and synthetic benchmarks (ModelNet40, ShapeNet) demonstrate that CA-Adv achieves a high attack success rate with a minimal perturbation budget. The generated adversarial examples maintain superior visual naturalness and geometric fidelity. The method provides a practical tool for evaluating the robustness of 3D recognition models in applications such as autonomous driving, urban-scale LiDAR perception, and remote sensing point cloud analysis. Full article

Background and Objectives: Sepsis and septic shock remain leading causes of morbidity and mortality in intensive care settings worldwide. While substantial epidemiological data exist from Western countries, the clinical profile of sepsis in regions with exceptionally high diabetes prevalence remains inadequately characterized. Saudi Arabia, with one of the highest diabetes mellitus prevalence rates globally, may exhibit distinct sepsis epidemiology, infection patterns, and outcomes. This study aimed to characterize the clinical profiles, antimicrobial management, and outcomes of sepsis and septic shock in a tertiary intensive care unit (ICU) in the Aseer region of southwestern Saudi Arabia. Materials and Methods: A retrospective observational study was conducted including 263 adults meeting Sepsis-3 criteria (232 sepsis, 31 septic shock) admitted to a tertiary ICU between January 2020 and December 2024. Demographics, comorbidities, laboratory parameters, microbiological data, antibiotic timing, interventions, and in-hospital mortality were analyzed. Logistic regression identified independent mortality predictors. This study adhered to the STROBE reporting guidelines. Results: The median age was 73 years with male predominance (58.4%). Diabetes mellitus (71.5%) and hypertension (65.8%) were highly prevalent. Urinary tract infections (UTIs) predominated (79.8%), with Escherichia coli as the most common pathogen (26.2%). The median time to antibiotic administration was 1.8 h; piperacillin–tazobactam was the most frequent empiric regimen (43.7%). Septic shock patients exhibited higher creatinine (1.65 vs. 1.08 mg/dL, p = 0.026) and lower platelets (194 vs. 271 × 10³/μL, p = 0.030). Mortality was 38.7% in septic shock versus 8.2% in sepsis (p < 0.001). Multivariate analysis confirmed septic shock (aOR: 5.23; 95% CI: 1.89–14.48) and mechanical ventilation (aOR: 15.42; 95% CI: 5.67–41.95) as independent mortality predictors. Conclusions: High diabetes prevalence shapes regional sepsis epidemiology with UTI predominance. Early antibiotic administration and recognition of septic shock remain critical for improving outcomes in this population. Full article

Recent advances in chiral analytical chemistry have revealed that fermented and natural foods contain substantial amounts of D-amino acids (D-AAs), the mirror-image counterparts of L-amino acids, leading to their recognition as nutraceutical components with potential health relevance. Although clinical evidence provides only limited support for their therapeutic efficacy, commercial expectations have outpaced scientific validation, and recent safety concerns emphasize the need for critical evaluation. In this review, we integrate findings from food chemistry, microbiology, biochemistry, physiology, and clinical research to provide a critical overview of dietary D-AAs. We examine how dietary exposure, microbial metabolism, host clearance capacity, and redox status collectively shape their context-dependent biological effects. We highlight the mechanistic linkage between D-amino acid oxidase (DAAO)-mediated hydrogen peroxide (H₂O₂) generation and organ-specific vulnerability, thereby clarifying the molecular basis of their “double-edged sword” actions. Within this interdisciplinary framework, we propose that D-AAs function as context-dependent “contronymic” molecules in cellular communication. By distinguishing physiological regulation, experimental modulation, and clinical application, this review aims to support evidence-based nutraceutical strategies and safety assessments that harness the potential benefits of D-AAs while minimizing associated risks. Full article

Study of early Chinese ceramics is crucial for understanding cultural, economic, and technological developments in Chinese history. With the evolving deep learning techniques, one urgent question would be, whether we can identify early Chinese ceramics by a simple 2D image without further domain knowledge. This work collected a highly diverse dataset for ancient Chinese ceramics from 15 dynasties, with 4 representative glaze colors and 15 shape types. We studied the performance of five state-of-the-art neural networks on two identification tasks: ceramic visual feature recognition and early Chinese ceramic dating. A class-imbalance learning strategy is designed to improve the models’ performance on multi-label tasks. To the best of our knowledge, our work is the first to introduce deep learning into early Chinese ceramic recognition on a large scale. Experiments prove that deep learning can recognize visual features like glaze and most shape types with high accuracy, while ceramic dating is feasible for the main dynasties but remains challenging along the overall history. Further quantitative assessment shows that cultural inheritance and artistic continuity can lead to reasonable false dating by classifying ceramics into adjacent dynasties or periods. Moreover, although domain knowledge is required for interpretation, deep learning shows great potential in recognizing even unlabeled time-relevant features, which can help study the inheritance and evolution of early Chinese ceramic development. Full article

Background/Objectives: The present study examined the effect of age on sentence recognition in noise in different noise conditions among adults with normal hearing sensitivity throughout the adult lifespan. Methods: A total of 113 adults aged between 21 and 65 years participated in the study; based on age, they were categorized into five groups. The sentence recognition was assessed in five noise conditions: speech-shaped noise (SSN), amplitude-modulated speech-shaped noise (AM-SSN), two-male-talker babble (2MB), four-male-talker babble (4MB), and four-female-talker babble (4FB). The sentences were presented at a signal-to-noise ratio of −5 dB in all noise conditions. Results: The sentence recognition scores declined with increasing age in all noise conditions. In addition, age had a differential effect on the sentence recognition scores in the AM-SSN and 2MB conditions compared with the SSN, 4MB, and 4FB conditions. In the AM-SSN and 2MB conditions, the scores were significantly different in the fourth decade compared with young adults. In other noises, the scores were significantly different after 30 years compared with younger adults. Further, across noise conditions, greater scores were obtained in the AM-SSN and 2MB conditions, and the lowest scores were obtained in the 4FB condition. Partial Spearman correlations revealed a moderate-to-strong negative correlation between age and sentence recognition scores across noise conditions. Conclusions: The findings of the present study showed that sentence recognition is negatively affected by age. In addition, age has a differential effect on sentence recognition in different noises. Full article

The rapid and accurate detection of SARS-CoV-2 biomarkers remains a critical requirement for effective outbreak control and decentralized diagnostics. Although RT-PCR is the current gold standard, its reliance on centralized laboratories and long processing times limits its applicability in point-of-care settings. In this context, optical biosensing platforms based on surface plasmon resonance (SPR) offer attractive features, including label-free, real-time, and quantitative detection. This study explores the use of synthetic receptors for the highly sensitive detection of the receptor-binding domain (RBD) of the SARS-CoV-2 spike protein. Specifically, soft molecularly imprinted polymer nanoparticles (nanoMIPs) were employed as synthetic receptors and integrated into a high-sensitivity, portable plasmonic platform based on a D-shaped plastic optical fiber (POF) SPR sensor. The nanoMIPs were selectively imprinted against the RBD, characterized by Dynamic Light Scattering (DLS), Isothermal Titration Calorimetry (ITC), and Scanning Electron Microscopy (SEM) to confirm nanoMIPs size, binding properties, and surface morphology. Next, the nanoMIPs were immobilized onto a gold-coated sensing surface, enabling enhanced specificity, affinity, and signal amplification compared to conventional biological recognition elements. The resulting RBD-SPR-nanoMIPs sensor demonstrated promising analytical performance, exhibiting high selectivity against potentially interfering proteins and an anticipated sensitivity suitable for RBD detection at femtomolar concentrations. The inherent stability of nanoMIPs suggests the potential for reusable SPR sensing platforms, paving the way for next-generation synthetic receptor-based plasmonic biosensors. Full article

The shipbuilding industry relies heavily on welding, which accounts for approximately 70% of the overall production process. However, the recent decline in skilled workers, together with rising labor costs, has accelerated the automation of shipbuilding operations. In particular, the welding activities are concentrated in the double-bottom region of ships, where collaborative robots are increasingly introduced to alleviate workforce shortages. Because these robots must directly recognize U-shaped weld joints, this study proposes an image-based classification system capable of automatically identifying and classifying such joints. In double-bottom structures, U-shaped weld joints can be categorized into 176 types according to combinations of collar plate type, slot, watertight feature, and girder. To distinguish these types, deep learning-based image recognition is employed. To construct a large-scale training dataset, 3D Computer-Aided Design (CAD) models were automatically generated using Open Cascade and subsequently rendered to produce synthetic images. Furthermore, to improve classification performance, the input images were split into left, right, upper, and lower regions for both training and inference. The class definitions for each region were simplified based on the presence or absence of key features. Consequently, the classification accuracy was significantly improved compared with an approach using non-split images. Full article

To address the limitation of 2D methods in inferring absolute scrap dimensions from images, we propose Scrap-SAM-CLIP (SSC), a vision-language model integrating the segment anything model (SAM) and contrastive language-image pre-training in Chinese (CN-CLIP). The model enables identification of canonical scrap shapes, establishing a foundational framework for subsequent 3D reconstruction and dimensional extraction within the 3D recognition pipeline. Individual modules of SSC are fine-tuned on the self-constructed scrap dataset. For segmentation, the combined box-and-point prompt yields optimal performance among various prompting strategies. MobileSAM and SAM-HQ-Tiny serve as effective lightweight alternatives for edge deployment. Fine-tuning the SAM decoder significantly enhances robustness under noisy prompts, improving accuracy by at least 5.55% with a five-positive-points prompt and up to 15.00% with a five-positive-points-and-five-negative-points prompt. In classification, SSC achieves 95.3% accuracy, outperforming Swin Transformer V2_base by 2.9%, with t-SNE visualizations confirming superior feature learning capability. The performance advantages of SSC stem from its modular assembly strategy, enabling component-specific optimization through subtask decoupling and enhancing system interpretability. This work refines the scrap 3D identification pipeline and demonstrates the efficacy of adapted foundation models in industrial vision systems. Full article

Background/Objectives: This review integrates evolutionary, metabolic, genetic, and nutritional perspectives to explain how sterol-derived vitamin D pathways shape human physiology and inter-individual variability in vitamin D status. Methods: The literature on sterol and vitamin D metabolism across animals, plants, fungi, and algae was synthesized with data from metabolomics databases, genome-wide association studies, RNA-seq resources (including GTEx), structural biology, and functional genomics. Results: Vitamin D₂ and vitamin D₃ likely emerged early in evolution as non-enzymatic photochemical sterol derivatives and were later co-opted into a tightly regulated endocrine system in vertebrates. In humans, cytochrome P450 enzymes coordinate vitamin D activation and degradation and intersect with oxysterol production, thereby linking vitamin D signaling to cholesterol and bile acid metabolism. Tissue-specific gene expression and regulatory genetic variants, particularly in the genes DHCR7, CYP2R1, CYP27B1, and CYP27A1, contribute to population-level differences in vitamin D status and metabolic outcomes. Structural analyses reveal selective, high-affinity binding of 1,25-dihydroxyvitamin D₃ to VDR, contrasted with broader, lower-affinity ligand recognition by LXRs. Dietary patterns modulate nuclear receptor signaling through distinct yet convergent ligand sources, including cholesterol-derived oxysterols, oxidized phytosterols, and vitamin D₂ versus vitamin D₃. Conclusions: Sterol and vitamin D metabolism constitute an evolutionarily conserved, adaptable network shaped by UV exposure, enzymatic control, genetic variation, and diet. This framework explains inter-individual variability in vitamin D biology and illustrates how evolutionary and dietary modulation of sterol-derived ligands confers functional flexibility to nuclear receptor signaling in human health. Full article

Microfluidic paper-based analytical devices (µPADs) convert ordinary cellulose into an active analytical platform where capillary gradients shape transport, surface chemistry guides recognition, and embedded electrodes or optical probes translate biochemical events into readable signals. Progress in fabrication—from wax and stencil barriers to laser-defined grooves, inkjet-printed conductive lattices, and 3D-structured multilayers—has expanded reaction capacity while preserving portability. Detection strategies span colorimetric fields that respond within porous fibers, fluorescence and ratiometric architectures tuned for low abundance biomarkers, and electrochemical interfaces resilient to turbidity, salinity, and biological noise. Applications now include diagnosing human body fluids, checking food safety, monitoring the environment, and testing for pesticides and illegal drugs, often in places with limited resources. Researchers are now using learning algorithms to read minute gradients or currents imperceptible to the human eye, effectively enhancing and assisting the measurement process. This perspective article focuses on the newest advancements in the design, fabrication, material selection, testing methods, and applications of µPADs, and it explains how they work, where they can be used, and what their future might hold. Full article

Every act of design tells a story about who belongs, who is seen, and who is heard. This paper looks at how participatory design-based research (PDR), practiced with relational care and reflexivity, can help shift interactions among marginalized youth from urban Indigenous and newcomer immigrant communities in Canada from othering toward understanding. Moving beyond surface-level celebrations of multiculturalism, the study frames design as a relational and ethical practice, one that surfaces assumptions, holds space for difference, and creates openings for intercultural dialogue. The study draws on a series of design circles (d.circles) in which youth co-created visual communication artefacts reflecting their lived experiences. These artefacts became catalysts for dialogue, enabling participants to challenge stereotypes, articulate concerns, and develop shared perspectives. Reflexivity was integral to the process, guiding both participants and the facilitator to consider power, positionality, and relational accountability throughout. Findings show that participatory design, grounded in Indigenous relational principles and participatory action research, can unsettle dominant narratives, foster mutual recognition, and support youth-led meaning-making. This work contributes to emerging conversations that position design thinking as a practice of ethical engagement rather than a tool for problem-solving alone. The learnings from this study show how critically practiced PDR can cultivate more inclusive and socially responsive pathways for intercultural understanding to take shape. Full article

Point cloud data, with its rich information and high-precision geometric details, holds significant value for urban road infrastructure surveying and management. To overcome the limitations of manual classification, this study employs deep learning techniques for automated point cloud feature extraction and classification, achieving high-precision object recognition in road scenes. By integrating the Princeton ModelNet40, ShapeNet, and Sydney Urban Objects datasets, we extracted 3D spatial coordinates from the Sydney Urban Objects Dataset and organized labeled point cloud files to build a comprehensive dataset reflecting real-world road scenarios. To address noise and occlusion-induced data gaps, three augmentation strategies were implemented: (1) Farthest Point Sampling (FPS): Preserves critical features while mitigating overfitting. (2) Random Z-axis rotation, translation, and scaling: Enhances model generalization. (3) Gaussian noise injection: Improves training sample realism. The PointNet++ framework was enhanced by integrating a point-filling method into the preprocessing module. Model training and prediction were conducted using its Multi-Scale Grouping (MSG) and Single-Scale Grouping (SSG) schemes. The model achieved an average training accuracy of 86.26% (peak single-instance accuracy: 98.54%; best category accuracy: 93.15%) and a test set accuracy of 97.41% (category accuracy: 84.50%). This study demonstrates successful road scene point cloud classification, providing valuable insights for point cloud data processing and related research. Full article

Precise 3D shape correspondence is a fundamental prerequisite for critical applications ranging from medical anatomical modeling to visual recognition. However, non-isometric 3D shape matching remains a challenging task due to the limited sensitivity of traditional Laplace–Beltrami (LB) bases to local geometric deformations such as stretching and bending. To address these limitations, this paper proposes a Sinkhorn-Regularized Elastic Functional Map framework (SRE-FMaps) that integrates entropy-regularized optimal transport with an elastic thin-shell energy basis. First, a sparse Sinkhorn transport plan is adopted to initialize a bijective correspondence with linear computational complexity. Then, a non-orthogonal elastic basis, derived from the Hessian of thin-shell deformation energy, is introduced to enhance high-frequency feature perception. Finally, correspondence stability is quantified through a cosine-based elastic distance metric, enabling retrieval and classification. Experiments on the SHREC2015, McGill, and Face datasets demonstrate that SRE-FMaps reduces the correspondence error by a maximum of 32% and achieves an average of 92.3% classification accuracy (with a peak of 94.74% on the Face dataset). Moreover, the framework exhibits superior robustness, yielding a recall of up to 91.67% and an F1-score of 0.94, effectively handling bending, stretching, and folding deformations compared with conventional LB-based functional map pipelines. The proposed framework provides a scalable solution for non-isometric shape correspondence in medical modeling, 3D reconstruction, and visual recognition. Full article

The growing frequency of extreme weather, earthquakes, fires, and environmental hazards underscores the need for real-time monitoring and predictive management at the urban scale. Conventional three-dimensional spatial information systems, which rely on orthophotos and ground surveys, often suffer from computational inefficiency and data overload when processing large and heterogeneous datasets. To address these limitations, this study introduces a three-dimensional GeoHash-based geocoding algorithm designed for lightweight, real-time, and attribute-driven digital twin operations. The proposed method comprises five integrated steps: generation of 3D GeoHash grids using longitude, latitude, and altitude coordinates; integration with GIS-based urban 3D models; level optimization using the Shape Overlap Ratio (SOR) with a threshold of 0.90; representative object labeling through weighted volume ratios; and altitude correction using DEM interpolation. Validation using a testbed in Sillim-dong, Seoul (10.19 km²), demonstrated that the framework achieved approximately 9.8 times faster 3D modeling performance than conventional orthophoto-based methods, while maintaining complete object recognition accuracy. The results confirm that the 3D GeoHash framework provides a unified spatial key structure that enhances data interoperability across querying, visualization, and simulation. This approach offers a practical foundation for operational digital twins, supporting high-efficiency 3D mapping and predictive disaster management toward resilient and data-driven urban systems. Full article

Traditional radar cross-section (RCS) prediction methods struggle with dynamically uncertain shapes of flexible targets, because they cannot disentangle intrinsic geometry from transient deformation, leading to degraded accuracy and prohibitive computational cost. To bridge this gap, we propose a dual-branch deep learning architecture that explicitly separates static geometric features from dynamic deformation characteristics, suppressing deformation noise in target identity representation. Training data are generated by coupling non-uniform rational B-spline (NURBS) parametric modeling with computational electromagnetics. The dynamic branch employs a one-dimensional convolutional neural network-long short-term memory-Transformer (1D-CNN-LSTM-Transformer) to extract temporal deformation features, while the static branch encodes baseline geometry via fully connected layers; their fused outputs deliver high-fidelity RCS predictions. Trained and tested on 1000 deformed metasurface samples, the proposed method achieves mean squared error (MSE) = 0.0541, root mean squared error (RMSE) = 0.2326 and coefficient of determination (R²) = 0.9997. The results demonstrate end-to-end accurate prediction under shape uncertainty, extending RCS modeling for flexible targets beyond recent studies that focus on static scenarios, and offering a reliable tool for flexible stealth design and high-resolution radar target recognition. Full article