Search Results (6,827)

Large volumes of images shared on social media have made image captioning an important tool for social hotspot identification and public opinion monitoring, where accurate visual–language alignment is essential for reliable analysis. However, existing image captioning models based on BLIP-2 (Bootstrapped Language–Image Pre-training) often struggle with complex, context-rich, and socially meaningful images in real-world social media scenarios, mainly due to insufficient cross-modal interaction, redundant visual token representations, and an inadequate ability to capture multi-scale semantic cues. As a result, the generated captions tend to be incomplete or less informative. To address these limitations, this paper proposes ECMA (Enhanced Cross-Modal Attention), a lightweight module integrated into the Querying Transformer (Q-Former) of BLIP-2. ECMA enhances cross-modal interaction through bidirectional attention between visual features and query tokens, enabling more effective information exchange, while a multi-scale visual aggregation strategy is introduced to model semantic representations at different levels of abstraction. In addition, a semantic residual gating mechanism is designed to suppress redundant information while preserving task-relevant features. ECMA can be seamlessly incorporated into BLIP-2 without modifying the original architecture or fine-tuning the vision encoder or the large language model, and is fully compatible with OPT (Open Pre-trained Transformer)-based variants. Experimental results on the COCO (Common Objects in Context) benchmark demonstrate consistent performance improvements, where ECMA improves the CIDEr (Consensus-based Image Description Evaluation) score from 144.6 to 146.8 and the BLEU-4 score from 42.5 to 43.9 on the OPT-6.7B model, corresponding to relative gains of 1.52% and 3.29%, respectively, while also achieving competitive METEOR (Metric for Evaluation of Translation with Explicit Ordering) scores. Further evaluations on social media datasets show that ECMA generates more coherent, context-aware, and socially informative captions, particularly for images involving complex interactions and socially meaningful scenes. Full article

The sustainable management of water resources and the development of climate-resilient infrastructure depend on the precise identification of water bodies in satellite imagery. This paper presents a novel deep learning architecture that integrates a convolutional block attention module (CBAM) into a modified EfficientNet–UNet backbone. This integration allows the model to prioritize informative features and spatial areas. The model robustness is ensured through a rigorous training regimen featuring five-fold cross-validation, dynamic test-time augmentation, and optimization with the Lovász loss function. The final model achieved the following values on the independent test set: precision = 90.67%, sensitivity = 86.96%, specificity = 96.18%, accuracy = 93.42%, Dice score = 88.78%, and IoU = 79.82%. These results demonstrate improvement over conventional segmentation pipelines, highlighting the effectiveness of attention mechanisms in extracting complex water-body patterns and boundaries. The key contributions of this paper include the following: (i) adaptation of CBAM within a UNet-style architecture tailored for remote sensing water-body extraction; (ii) a rigorous ablation study detailing the incremental impact of decoder complexity, attention integration, and loss function choice; and (iii) validation of a high-fidelity, computationally efficient model ready for deployment in large-scale water-resource and ecosystem-monitoring systems. Our findings show that attention-guided segmentation networks provide a robust pathway toward high-fidelity and sustainable water-body mapping. Full article

Background/Objectives: Safe delivery and correct identification of newborns are critical aspects of healthcare systems globally. The accreditation of healthcare and standards regulation significantly promotes the adoption of modern technologies to address risks related to infant abduction and misidentification. The effectiveness and extent of these mandates vary across settings and countries. Therefore, this study aims to map and explore modern technologies used for safe newborn delivery and correct identification aligned with healthcare accreditation and regulatory frameworks. Methods: This review adheres to the Preferred Reporting Items for Systematic Review and Meta-Analysis extension for scoping reviews (PRISMA-ScR) guidelines. The Problem, Intervention, Comparison, and Outcome (PICO) framework was employed to facilitate the development of the research question. This study examined studies reporting technologies such as radio frequency identification (RFID), biometric identification, and real-time monitoring across healthcare settings for infant protection through the Normalization Process Theory (NPT). Among three databases and search engines (PubMed, Google Scholar, and Web of Science). The risk of bias for each study was assessed using the AACODS Checklist, SQUIRE 2.0 Checklist, TIDieR Checklist, and JBI tools. Results: Out of 8753 records, only 27 reports were eligible to be included in this review. The most frequently reported technologies were RFID systems (11 studies, 37.9%) and biometric systems such as footprint and facial recognition (6 studies, 20.7%). Despite strong technological potential, many healthcare institutions struggled with the adoption of infant protection technologies. Accreditation systems among the high-resource settings actively mandate advanced technologies and support the integration of staff training and simulation drills. Comparably, middle- and low-income regions usually face challenges related to regulatory enforcement, infrastructure, staff readiness, and limited adoption of modern technologies. Conclusions: Accreditation and standards development are critical catalysts for the adoption of modern infant protection technology. Standards must be comprehensible, adaptable, and supported by investment in human resources and infrastructure. Future regulation must focus on strengthening enforcement, continuous quality improvement, and capacity building to achieve sustainable protection across the world. Full article

Effective and sustainable livestock management within Mediterranean ecosystems depends heavily on accurate and timely monitoring of sward composition. Traditionally, this task has relied on human observers who must traverse large and often rugged areas to identify the distribution of grasses, legumes, shrubs, and other plant groups. However, this approach is not only labor-intensive and slow but also susceptible to substantial human error, especially when observations must be repeated frequently or carried out under difficult field conditions. In the present study, an alternative method that integrates wearable cameras with modern computer-vision techniques to automatically recognize pasture plant species through an edge device present in farm premises was investigated. Additionally, the feasibility of achieving reliable classification performance on resource-constrained edge devices was evaluated. To this end, five widely used pre-trained convolutional neural networks were compared against a lightweight custom model developed entirely from scratch. The results demonstrated that ResNet50 delivered the strongest classification accuracy, achieving a Matthews Correlation Coefficient (MCC) of 0.992. Nonetheless, the custom lightweight model proved to be a practical compromise for real-world field use, reaching an MCC of 0.893 while requiring only 6.24 MB of storage. The inference performance on Raspberry Pi 4, Raspberry Pi 5, and Jetson Orin Nano platforms was also evaluated, revealing that the Selective Search stage remains a major computational limitation for achieving real-time operation. The results obtained confirm the possibility of implementing a plant identification system in agricultural facilities without the need to transfer images to a cloud-based application. Full article

Ship-to-ship (STS) transfer has become a major concern for maritime security and regulatory authorities, as it is frequently exploited for smuggling and other illicit activities. Accurate and timely identification of STS events is therefore essential for effective maritime supervision. Existing monitoring approaches, however, suffer from two inherent limitations: AIS-based surveillance is vulnerable to intentional signal shutdown or manipulation, and remote-sensing-based ship detection alone lacks digital identity information and cannot assess the legitimacy of transfer activities. To address these challenges, we propose a Multi-source Observation framework with SAR and AIS (MOSA), which integrates SAR imagery with AIS data. The framework consists of two key components: STS-YOLO, a high-precision fine-grained ship detection model, in which a dynamic adaptive feature extraction (DAFE) module and a multi-attention mechanism (MAM) are introduced to enhance feature representation and robustness in complex maritime SAR scenes, and the SAR-AIS Consistency Analysis Workflow (SACA-Workflow), designed to identify suspected abnormal STS behaviors by analyzing inconsistencies between physical and digital ship identities. Experimental results on the SDFSD-v1.5 dataset demonstrate the quantitative performance gains and improved fine-grained detection performance of STS-YOLO in terms of standard detection metrics. In addition, generalization experiments conducted on large-scene SAR imagery from the waters near Panama and Singapore, in addition to multi-satellite SAR data (Capella Space and Umbra) from the Gibraltar region, validate the cross-regional and cross-sensor robustness of the proposed framework. The effectiveness of the SACA-Workflow is evaluated qualitatively through representative case studies. In all evaluated scenarios, the SACA-Workflow effectively assists in identifying suspected abnormal STS events and revealing potential AIS inconsistency indicators. Overall, MOSA provides a robust and practical solution for multi-scenario maritime monitoring and supports reliable detection of suspected abnormal STS activities. Full article

Background/Objectives: The study primarily examined whether an IoT-based medication assistance system enhances ART adherence relative to standard care, and secondarily evaluated device feasibility and error patterns over time. Methods: This prospective study was conducted between June 2022 and October 2023 at a tertiary hospital in South Korea. Adults (≥19 years) living with HIV and prescribed ART were included; those with comorbid hepatitis B or C were excluded. People living with HIV who agreed to use the IoT-based InPHRPILL system (Sofnet Inc., Seoul, Republic of Korea) were assigned to the intervention group, whereas those who declined were assigned to the control group. Viral suppression, CD4⁺ cell counts, and adherence rates were measured. Additional analyses evaluated 12-month longitudinal adherence using pill-count data in both groups, and device-measured adherence and device-associated error rates in the intervention group. Results: Thirty-five participants (12 in the intervention group and 23 in the control group) were included. The intervention group demonstrated marginally shorter durations since HIV diagnosis and ART initiation at study enrollment, as well as slightly higher baseline HIV-RNA levels; however, these differences did not reach statistical significance. The median pill-counting and IoT device adherence rates were 100% and 87.4%, respectively (median deviation error rate = 4.4%). Poisson regression revealed significantly reduced error rates over time (β = −0.06493, p < 0.01), suggesting improved device use proficiency. Conclusions: IoT-based medication assistance systems may provide objective, real-time monitoring of ART adherence and facilitate identification of discrepancies between clinical evaluations and actual adherence patterns. Larger studies targeting individuals with suboptimal adherence are warranted to determine whether such systems can enhance adherence outcomes. Full article

With the increasing deployment of grid-forming static var generators (GFM-SVG) in modern power systems, the reliability of the valve hall that houses the core power modules has become a critical concern. To overcome the limitations of conventional wired monitoring systems—complex cabling, poor scalability, and incomplete state perception—this paper proposes and implements a multi-source fusion wireless data acquisition system specifically designed for GFM-SVG valve halls. The system integrates acoustic, visual, and infrared sensing nodes into a wireless sensor network (WSN) to cooperatively capture thermoacoustic visual multi-physics information of key components. A dual-mode communication scheme, using Wireless Fidelity (Wi-Fi) as the primary link and Fourth-Generation Mobile Communication Network (4G) as a backup channel, is adopted together with data encryption, automatic reconnection, and retransmission-checking mechanisms to ensure reliable operation in strong electromagnetic interference environments. The main innovation lies in a multi-source information fusion algorithm based on an improved Dempster–Shafer (D–S) evidence theory, which is combined with the object detection capability of the You Only Look Once, Version 8 (YOLOv8) model to effectively handle the uncertainty and conflict of heterogeneous data sources. This enables accurate identification and early warning of multiple types of faults, including local overheating, abnormal acoustic signatures, and coolant leakage. Experimental results demonstrate that the proposed system achieves a fault-diagnosis accuracy of 98.5%, significantly outperforming single-sensor approaches, and thus provides an efficient and intelligent operation-and-maintenance solution for ensuring the safe and stable operation of GFM-SVG equipment. Full article

The preservation of cultural heritage within museum environments requires systematic control and monitoring of indoor microclimatic conditions. Over the past four decades, scientific evidence has established the critical role of environmental parameters, including air temperature, relative humidity, light, and airborne pollutants, in the preventive conservation of artifacts. International standards and national guidelines mandate continuous, non-invasive monitoring protocols that integrate conservation requirements with the architectural and operational constraints of historic buildings. Effective implementation necessitates a multidisciplinary approach balancing artifact preservation, human comfort, and building energy efficiency. Recent international recommendations further promote adaptive approaches wherein microclimate thresholds are calibrated to site-specific “historical climate” conditions, derived from minimum one-year baseline datasets. While essential for long-term conservation management, the design and implementation of such monitoring systems present significant technical and logistical challenges. This study presents a replicable methodological approach wherein preliminary surveys and three short-term monitoring campaigns (duration: 2 to 5 weeks) supported design, sensor selection, and spatial deployment and will allow the validation of a long-term continuous monitoring infrastructure (at least one year). These preliminary investigations enabled the following: (1) identification of priority environmental parameters; (2) optimization of sensor placement relative to exhibition layouts and maintenance protocols; and (3) preliminary assessment of microclimate risks in naturally ventilated spaces in the absence of HVAC systems. Full article

Sepsis remains a leading cause of global mortality and is characterized by a dysregulated host immune response to infection. Early deaths often result from hyperinflammation and organ dysfunction, whereas late-stage mortality is increasingly attributed to sepsis-induced immunosuppression, leading to secondary infections and viral reactivation. Challenges persist in the identification and management of sepsis-induced immunosuppression, including the lack of standardized immune monitoring methods, the absence of reliable immune biomarkers to guide therapy, and the limited success of immunomodulatory therapies in clinical trials. This review comprehensively summarizes the pathophysiology of sepsis-induced immunosuppression, encompassing immune cell apoptosis and exhaustion, the expansion and activation of immunomodulatory cells, metabolic reprogramming, epigenetic alterations, and iatrogenic factors. We also discuss current diagnostic challenges and explore emerging immunomodulatory strategies, such as cytokine therapies, immune checkpoint inhibitors, and metabolic modulators, as potential approaches to restore immune function. Finally, we highlight the importance of immune phenotyping and individualized precision medicine in the future management of sepsis, and integrating multidisciplinary approaches from mechanistic research to targeted therapies holds promise for improving patient outcomes. Full article

Background: Salt-sensitive blood pressure (SSBP) represents a prevalent yet underrecognized hypertensive phenotype, in which blood pressure (BP) and volume status are disproportionately influenced by dietary sodium intake. Beyond BP elevation alone, salt sensitivity reflects a convergence of renal sodium handling abnormalities, neurohormonal activation, vascular dysfunction, and inflammatory pathways that link excessive sodium exposure to progressive kidney injury and adverse cardiac remodeling. Given its association with chronic kidney disease (CKD) and the association of heart failure with preserved ejection fraction (HFpEF), improved recognition of SSBP has direct clinical relevance. Objective: This narrative review aims to synthesize current mechanistic and clinical evidence on SSBP, focusing on pathophysiology, cardiorenal interactions, diagnostic challenges, and phenotype-guided therapeutic strategies with practical applicability. Methods: A narrative literature review was conducted using PubMed, Scopus, and Web of Science from inception through January 2026. Experimental, translational, and clinical studies, along with relevant guideline documents, were integrated to provide conceptual and clinical interpretation rather than quantitative analysis. Key Findings: Impaired renal sodium excretion, intrarenal RAAS activation, sympathetic overactivity, endothelial dysfunction, and immune-mediated inflammation contribute to sodium retention, microvascular dysfunction, and fibrotic remodeling across the kidney–heart axis. These pathways are strongly supported by experimental and translational data, but direct interventional clinical validation remains limited for several mechanisms. Clinically, salt-sensitive individuals often exhibit non-dipping BP patterns, albuminuria, salt-induced edema, and a predisposition to HFpEF. Dynamic BP monitoring combined with targeted laboratory assessment improves identification of this phenotype and supports individualized management. Conclusions: Early recognition of SSBP enables targeted interventions beyond uniform sodium restriction. Phenotype-guided strategies integrating lifestyle modification, RAAS blockade, thiazide-like diuretics, mineralocorticoid receptor antagonists, and sodium-glucose co-transporters 2 inhibitors (SGLT2i) may improve cardiorenal outcomes. Emerging precision tools (e.g., wearable blood-pressure sensors, digital sodium tracking technologies, etc.) remain exploratory but may further refine individualized management. Full article

Structural Health Monitoring (SHM) of existing bridges increasingly relies on dynamic measurements to assess structural performance and detect potential damage. However, the practical implementation of long-term vibration-based monitoring is still constrained by the volume of data required and the complexity of continuous acquisition systems. In the context of ensuring the safety and performance of existing bridge infrastructure, vibration-based monitoring offers a powerful tool for detecting changes in structural behavior. This study presents an extended investigation of dynamic monitoring applied to composite steel–concrete viaducts, focusing particularly on the signal-analysis framework and methodological enhancements. Short-duration accelerometric records are processed through an automated signal-selection pipeline and advanced modal-parameter extraction algorithms to yield identification of modal features. Emphasis is placed on the statistical evaluation of modal-parameter stability, effects of operational and environmental variability, and the potential for long-term trend detection. The results highlight the limits of short-length recordings when OMA techniques are applied. Nevertheless, appropriate signal processing and data handling can provide acceptable insights into the dynamic characteristics of large bridge systems. The methodological findings provide a foundation for improved monitoring workflows, showing the amount of information that can be retrieved using a cost-effective hardware deployment and supporting further development toward structural digital twins. Full article

The depth of corrosion pits is critical to the safe operation of pipelines. Conventionally, monitoring of metal pipeline corrosion has generally relied on a single technique. Among these, the ultrasonic thickness measurement method and the field signature method are widely used in pipeline monitoring systems. The ultrasonic method has high reliability and accuracy for monitoring the thickness of pipelines, but it is limited by coupling and measurement range. Additionally, the field signature method often suffers from inadequate identification of corrosion pits and lower measurement accuracy. To address these limitations, a multi-sensor fusion model is proposed to monitor corrosion in metal pipelines. The multi-sensor fusion model is constructed by alternately arranging ultrasonic sensors and field signature probes, and a dedicated fusion algorithm is designed. The integrated model leverages the complementary strengths of both techniques while mitigating their individual shortcomings. Furthermore, an artificial neural network is employed to accurately identify pitting depth, thereby resolving the challenge in discriminating corrosion pit depths. Experimental results demonstrate that the multi-sensor fusion model can overcome the inherent drawbacks associated with a single technique. Consequently, it enhances the overall reliability, measurement accuracy, and operational range of the pipeline corrosion measurement system. Full article

Accurate, real-time vehicle identification is essential for data-driven urban planning, enabling applications like emissions monitoring and the enforcement of environmental regulations. However, identifying a vehicle’s model, generation, and production year remains a significant challenge for VMMR systems. This is especially true when cameras capture multiple vehicles simultaneously under suboptimal imaging conditions. This challenge is amplified in Europe, as most existing VMMR datasets are designed for non-European markets. To address this, we present two contributions: a newly curated dataset of 84,732 images across 625 classes, and a robust two-stage YOLO11 system trained on this data. The dataset focuses on the European market and realistic viewpoints like front and rear angles. The system, comprising a Vehicle Localization Module (VLM) and a Fine-Grained Classification Module (FGCM), performs detailed model classification without relying on license plates or additional sensors. When tested on real European traffic footage, our system achieved 80% accuracy and outperformed models trained on U.S.-centric datasets. Full article

Background/Objectives: Necrotizing enterocolitis (NEC) is one of the most devastating gastrointestinal emergencies in premature neonates, with particularly high mortality among those requiring surgical intervention. Early identification of high-risk patients remains challenging. This study aimed to evaluate the prognostic value of complete blood count-derived inflammatory indices for predicting mortality in premature infants undergoing surgery for NEC. Methods: A total of 74 premature neonates with Bell stage II or III NEC who underwent surgical treatment between 2018 and 2023 were retrospectively analyzed. Preoperative and postoperative hematologic inflammatory indices, including neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), systemic immune-inflammation index (SII), and platelet-to-neutrophil ratio (PNR), were recorded. Receiver operating characteristic (ROC) curve analysis was used to assess predictive performance. Variables with p < 0.10 in univariate analysis were entered into multivariate logistic regression models. Results: Overall mortality was 35.1%. Non-survivors had significantly lower gestational age and birth weight and a higher prevalence of advanced disease. Preoperatively, NLR was higher and PNR was lower in non-survivors. Postoperatively, NLR and C-reactive protein levels increased, while PNR showed a marked decline in infants who died. ROC analysis identified postoperative PNR as the strongest predictor of mortality, followed by preoperative SII and postoperative NLR. Multivariate analysis demonstrated that lower gestational age, advanced disease stage, and reduced postoperative PNR were independently associated factors for mortality. Conclusions: Postoperative reduction in platelet-to-neutrophil ratio appears to be a practical, inexpensive, and easily obtainable biomarker for early risk stratification in surgically treated NEC. Incorporating routine hematologic inflammatory indices into postoperative monitoring may support timely identification of high-risk infants and guide individualized clinical management in neonatal intensive care units. Full article

Climate change is rapidly altering the distribution and abundance of species, with significant impacts on regional ecosystems, including reduced ecosystem services and the loss of biodiversity. Accurately predicting changes in the distribution and abundance of taxa under future climate scenarios is, therefore, crucial. In South Korea, beetle data collected via pitfall traps from approximately 300 forest sites between 2007 and 2009 (30 families, 4 genera, and 150 species) were used to forecast changes in their abundance and distribution under climate change scenarios RCP 4.5 and 8.5. This study evaluated the accuracy of those predictions using data from a subsequent survey conducted between 2017 and 2019. We compared species richness, abundance, changes in abundance (i.e., number of individuals), and occurrence (i.e., number of occupied sites) using data from 273 sites that were surveyed in both the initial (2007–2009) and follow-up (2017–2019) periods. All four parameters were found to be significantly influenced by the identifiers. This identifier bias was attributed to the omission of morphologically similar species in the initial survey or the loss of individuals during the preparation process of dry specimens. As a result, increases in abundance and distribution appear to have been affected by identification errors, whereas decreases more closely reflect actual ecological changes. When the comparison between predicted and observed results was restricted to taxa with reduced abundance and distribution, the number of taxa that matched the predictions was significantly higher than that of those that did not. Based on ease of identification, abundance, and sensitivity to climate change, we selected a set of indicator taxa (four families, two genera, and seven species) for climate change monitoring. Full article