Search Results (90)

Industrial image anomaly detection requires the simultaneous identification of local structural and global logical anomalies. Existing methods specialize in single-type anomalies due to divergent feature requirements: structural anomalies demand fine-grained local features, while logical anomalies need semantic features. Consequently, designing a unified network architecture that effectively captures both features without task conflicts remains a key challenge. To address this problem, we propose a Dual-Stream Memory Bank Anomaly Detection (DSMBAD) framework, which enables the collaborative detection of both structural and logical anomalies from complementary perspectives. The framework consists of two memory banks: one stores patch features for detecting structural anomalies through local feature discrepancies, while the other uses segmentation maps to model component relationships for logical anomaly identification. Additionally, a feature distillation mechanism aligns features from different backbone networks to enhance global semantic information. We also introduce a shape-based anomaly scoring method that quantifies differences in component relationships using spatial–morphological features. Experimental results on the MVTec LOCO AD dataset show that our method achieves 91.0% I-AUROC (logical) and 90.8% (structural), significantly outperforming single-type models. Ablation studies confirm the dual-stream design and module effectiveness, offering a novel unified solution. Full article

Adequate and diverse diets are essential for children’s physical and cognitive development, yet food insecurity and malnutrition continue to threaten this fundamental right, which remains a pressing concern in many resource-poor settings. This study investigated food and nutrition security in Early Childhood Development (ECD) centres in Makhanda, South Africa, through a community-based participatory research approach. Using a mixed-methods approach combining questionnaire interviews, focus group discussions, direct observations, and community asset mapping across eight ECD centres enrolling 307 children aged 0–5 years, the study engaged ECD facilitators and analysed dietary practices across these centres. Results indicated that financial constraints severely affect the quality and diversity of food provided at the centres, thus undermining the ability to provide nutritionally adequate meals. The average amount spent on food per child per month at the centres was R90 ± R25 (South African Rand). Although three meals were generally offered daily, cost-driven dietary substitutions with cheaper, less diverse alternatives, often at the expense of nutritional value, were common. Despite guidance from Department of Health dieticians, financial limitations contributed to suboptimal feeding practices, with diets dominated by grains and starchy foods, with limited access to and rare consumption of protein-rich foods, dairy, and vitamin A-rich fruits and vegetables. ECD facilitators noted insufficient parental contributions and low engagement in supporting centre operations and child nutrition provision, indicating a gap in awareness and limited nutrition knowledge regarding optimal infant and young child feeding (IYCF) practices. The findings emphasise the need for sustainable, multi-level and community-led interventions, including food gardening, creating ECD centre food banks, parental nutrition education programmes, and enhanced financial literacy among ECD facilitators. Strengthening local food systems and establishing collaborative partnerships with communities and policymakers are essential to improve the nutritional environment in ECD settings. Similarly, enhanced government support mechanisms and policy-level reforms are critical to ensure that children in resource-poor areas receive adequate nutrition. Future research should focus on scalable, locally anchored models for sustainable child nutrition interventions that are contextually grounded, community-driven, and should strengthen the resilience of ECD centres in South Africa. Full article

Through laboratory experiments in an S-shaped channel, this study analyzes how the flow Froude number, the ratio of ice-to-flow rate, pier spacing-diameter ratio, and bed material median grain size influence scour depth around side-by-side double piers under ice-jammed flow conditions. Unlike the development of a scour hole around a bridge pier in a straight channel, where the scour depth increases with the flow Froude number under ice-covered conditions, this study reveals that in an S-shaped channel, scour depth increases with the flow Froude number near the convex bank pier and decreases near the concave bank counterpart. Irrespective of ice conditions, a higher ratio of pier spacing-diameter correlates with augmented scour depth at the convex bank and diminished scour at the concave bank. As the ice-to-flow rate ratio increases, the ice jam thickness in the S-shaped channel also increases, leading to a significant decrease in the flow area and resulting in deeper scour holes around the piers. Equations have been developed to calculate the maximum scour depth around side-by-side double piers positioned in an S-shaped channel with ice-jammed flow. Full article

Estimating bed load in rivers is a critical aspect of river engineering. Numerous methods have been developed to quantify bed load transport, often yielding varying results depending on the bed surface texture and grain size. This study aims to investigate how vegetation on channel banks and bed material particle size influence bed load transport, bed shear stress, velocity distribution, and the Shields parameter. It also examines the impact of geometric changes in the channel cross-section on bed load transport capacity. To address these objectives, a novel simulation method was developed to analyze the effects of vegetated banks, bed material size, and channel geometry. Field investigations were carried out in two reaches of the Taleghan River in Iran—one with vegetated banks and one without. Complementary flume experiments were conducted at two scales, incorporating vegetation on the sidewalls. Results showed that Shields parameter distribution corresponded with bed load distribution across cross-sections. Increase in flow rate and the Shields parameter led to higher bedload transport rates. Near vegetated banks, flow velocity, shear stress, and bedload transport were significantly reduced, with velocity profiles showing distinct variations compared to non-vegetated sections. Full article

The EAT-Lancet diet outlines target consumption for specific food components but overlooks accessibility and cost issues, which may hinder adherence among vulnerable populations. This study examines the acquisition profile of EAT-Lancet diet components by food security and poverty status, using data from 57,920 households in the 2017–2018 Brazilian Household Budget Survey. Poverty and food insecurity were defined according to the World Bank per capita income cutoffs and the Brazilian Food Insecurity Scale, respectively. Food acquisition was classified into 15 EAT-Lancet diet components and expressed as per capita daily averages (g, % of total available energy, and % of food expenditure), by food security and poverty strata. Brazilian households were 37.9% food-insecure and 12% poor. Compared to more privileged counterparts, these households prioritized the acquisition of staples like refined cereals and legumes over most EAT-Lancet diet adequacy components, such as fruits, vegetables, whole grains, nuts, and peanuts. While lower energy shares from moderation components were only slightly evident for red meat and dairy among food-insecure households, pronounced reductions in added sugars and vegetable oils were seen among the poor. These findings suggest that public policies should synergically address particularities of different deprivation contexts to promote sustainable diets in Brazil. Full article

Object detection in unmanned aerial vehicle (UAV) images poses significant challenges due to complex scale variations and class imbalance among objects. Existing methods often address these challenges separately, overlooking the intricate nature of UAV images and the potential synergy between them. In response, this paper proposes AD-Det, a novel framework employing a coherent coarse-to-fine strategy that seamlessly integrates two pivotal components: adaptive small object enhancement (ASOE) and dynamic class-balanced copy–paste (DCC). ASOE utilizes a high-resolution feature map to identify and cluster regions containing small objects. These regions are subsequently enlarged and processed by a fine-grained detector. On the other hand, DCC conducts object-level resampling by dynamically pasting tail classes around the cluster centers obtained by ASOE, maintaining a dynamic memory bank for each tail class. This approach enables AD-Det to not only extract regions with small objects for precise detection but also dynamically perform reasonable resampling for tail-class objects. Consequently, AD-Det enhances the overall detection performance by addressing the challenges of scale variations and class imbalance in UAV images through a synergistic and adaptive framework. We extensively evaluate our approach on two public datasets, i.e., VisDrone and UAVDT, and demonstrate that AD-Det significantly outperforms existing competitive alternatives. Notably, AD-Det achieves a $37.5\%$ average precision (AP) on the VisDrone dataset, surpassing its counterparts by at least $3.1\%$. Full article

In today’s digital economy, long-term business success increasingly depends on both financial resources and digital capabilities. However, limited research explores how these two factors jointly drive sustainable performance in SMEs. This study investigates how access to finance influences sustainability outcomes among SMEs, with digital agility as a mediator and Industry 5.0 as a moderator. Based on cross-sectional data collected from 383 Hungarian manufacturing SMEs in late 2024, we apply PLS-SEM and Machine Learning (ML) techniques to validate our model. The results show that access to finance significantly influences digital agility and SMEs’ sustainability. Digital agility significantly mediates between access to finance and SMEs’ sustainability. Industry 5.0 further strengthens the relationships between access to finance and both SMEs’ sustainability and digital agility. ML identified digital agility as the key factor of SMEs’ sustainability. This study contributes to the Resource-Based View and Triple Bottom Line views by synergizing digital agility and human-centered Industry 5.0. Theoretically, it also supports methodological innovation in showing that the combined usage of PLS-SEM and ML can produce stronger and more fine-grained conclusions on complex sustainability dynamics. The findings are practically relevant guidance for SMEs, policymakers, and banks intending to enable digitally facilitated sustainable growth. To the scientific community, this study bridges a critical void by linking finance, technology, and sustainability within an innovative framework. Socially, it highlights how SMEs’ financial and digital capabilities can be strengthened not only to drive economic performance but also to support environmental sustainability and social well-being—resulting in inclusive and sustainable growth for emerging economies. Full article

Bone grafting is one of the most used surgical techniques to favor bone regeneration and repair in orthopedic procedures. Despite autografting continuing to be considered the gold standard, allogeneic bone tissues remain a viable alternative albeit in the last decades, only a few studies have been carried out to translate enhanced allogeneic bone grafts into clinical solutions. In this in vitro study, cortical allogeneic bone samples were coated with copper-doped bioactive glass 45S5 (Cu-BG) by means of the pulsed-laser deposition technique to combine the mechanical support and osteoconductive properties of human bone with the osteogenic and pro-angiogenic features of the bioactive material. Contact angle (CA), scanning electron microscopy (SEM), and atomic force microscopy (AFM) measurements were carried out to quantitatively compare the impact on the bone surface properties of the morphological changes induced by the presence of the coating. Specifically, the obtained results have shown a total absorption of the drop on the coated samples. The coating on the bone tissue surface consisted of a homogeneous Cu-BG background layer with micrometric grain-like aggregates on it—a morphological feature that can facilitate osteoblast adhesion and proliferation. Cytotoxicity and cell viability were carried out on Saos-2 osteoblast-like cells, demonstrating the biocompatibility of the novel composite bone tissue and the absence of cytotoxic residuals. Moreover, human bone marrow-derived mesenchymal stem cells (hBMSCs) were seeded on Cu-BG and not-coated (NC) samples to evaluate the bioactivity and their differentiation toward the osteogenic phenotype. Our findings showed the pro-osteogenic and pro-angiogenic potential of Cu-BG coatings, although dynamic changes were observed over time. At seven days, the Cu-BG samples exhibited significantly higher expressions of SP7, SPP1, and BGLAP genes, indicating an enhanced early osteogenic commitment. Moreover, VEGF expression was significantly increased in Cu-BG compared to the control. These results pave the way for the development of an innovative class of bone-based products distributed by tissue banks. Full article

This comprehensive survey explores recent advancements in scheduling techniques for efficient deep learning computations on GPUs. The article highlights challenges related to parallel thread execution, resource utilization, and memory latency in GPUs, which can lead to suboptimal performance. The surveyed research focuses on novel scheduling policies to improve memory latency tolerance, exploit parallelism, and enhance GPU resource utilization. Additionally, it explores the integration of prefetching mechanisms, fine-grained warp scheduling, and warp switching strategies to optimize deep learning computations. These techniques demonstrate significant improvements in throughput, memory bank parallelism, and latency reduction. The insights gained from this survey can guide researchers, system designers, and practitioners in developing more efficient and powerful deep learning systems on GPUs. Furthermore, potential future research directions include advanced scheduling techniques, energy efficiency considerations, and the integration of emerging computing technologies. Through continuous advancement in scheduling techniques, the full potential of GPUs can be unlocked for a wide range of applications and domains, including GPU-accelerated deep learning, task scheduling, resource management, memory optimization, and more. Full article

Semantic segmentation and height estimation tasks in remote sensing imagery exhibit distinctive characteristics, including scale sensitivity, category imbalance, and insufficient fine details. Recent approaches have leveraged multi-task learning methods to jointly predict these tasks along with auxiliary tasks, such as edge detection, to improve the accuracy of fine-grained details. However, most approaches only acquire knowledge from auxiliary tasks, disregarding the inter-task knowledge guidance across all tasks. To address these challenges, we propose KMNet, a novel architecture referred to as a knowledge-guided multi-task network, which can be applied to different primary and auxiliary task combinations. KMNet employs a multi-scale methodology to extract feature information from the input image. Subsequently, the architecture incorporates the multi-scale knowledge-guided fusion (MKF) module, which is designed to generate a comprehensive knowledge bank serving as a resource for guiding the feature fusion process. The knowledge-guided fusion feature is then utilized to generate the final predictions for the primary tasks. Comprehensive experiments conducted on two publicly available remote sensing datasets, namely the Potsdam dataset and the Vaihingen dataset, demonstrate the effectiveness of the proposed method in achieving impressive performance on both semantic segmentation and height estimation tasks. Codes, pre-trained models, and more results will be publicly available. Full article

The objective of this study was to evaluate the genetic diversity among traditional common bean accessions through morphological descriptors and molecular markers. Sixty-seven common bean accessions from the Germplasm bank of the Instituto Federal of Espírito Santo—Campus de Alegre were evaluated. For this, 25 specific morphological descriptors were used, namely 12 quantitative and 13 qualitative ones. A diversity analysis based on morphological descriptors was carried out using the Gower algorithm. For molecular characterization, 23 ISSR primers were used to estimate dissimilarity using the Jaccard Index. Based on the dendrograms obtained by the UPGMA method, for morphological and molecular characterization, high genetic variability was observed between the common bean genotypes studied, evidenced by cophenetic correlation values in the order of 0.99, indicating an accurate representation of the dissimilarity matrix by the UPGMA clustering. In the morphological characterization, high phenotypic diversity was observed between the accessions, with grains of different shapes, colors, and sizes, and the accessions were grouped into nine distinct groups. Molecular characterization was efficient in separating the genotypes in the Andean and Mesoamerican groups, with the 23 ISSR primers studied generating an average of 6.35 polymorphic bands. The work identified divergent accessions that can serve different market niches, which can be indicated as parents to form breeding programs in order to obtain progenies with high genetic variability. Full article

Barley is one of the most important cereal crops in the world, and its value as a food is constantly being revealed, so the research into and the use of barley germplasm are very important for global food security. Although a large number of barley germplasm samples have been collected globally, their specific genetic compositions are not well understood, and in many cases their origins are even disputed. In this study, 183 barley germplasm samples from the Shanghai Agricultural Gene Bank were genotyped using genotyping-by-sequencing (GBS) technology, SNPs were identified and their genetic parameters were estimated, principal component analysis (PCA) was preformed, and the phylogenetic tree and population structure of the samples were also analyzed. In addition, a genome-wide association study (GWAS) was carried out for the hulled/naked grain trait, and a KASP marker was developed using an associated SNP. The results showed that a total of 181,906 SNPs were identified, and these barley germplasm samples could be roughly divided into three categories according to the phylogenetic analysis, which was generally consistent with the classification of the traits of row type and hulled/naked grain. Population structure analysis showed that the whole barley population could be divided into four sub-populations (SPs), the main difference from previous classifications being that the two-rowed and the hulled genotypes were sub-divided into two SPs. The GWAS analysis of the hulled/naked trait showed that many associated loci were unrelated to the Nud/nud locus, indicating that there might be new loci controlling the trait. A KASP marker was developed for one exon-type SNP on chromosome 7. Genotyping based on the KASP assay was consistent with that based on SNPs, indicating that the gene of this locus might be associated with the hulled/naked trait. The above work not only lays a good foundation for the future utilization of this barley germplasm population but it provides new loci and candidate genes for the hulled/naked trait. Full article

(1) Background: Fortifying maize and wheat flours with folic acid has effectively reduced neural tube defect-affected births. However, maize and wheat flours may not be widely consumed in all countries; further reduction in neural tube defect-affected births could benefit from the identification of alternative food vehicles. We aimed to use dietary intake or apparent consumption data to determine alternative food vehicles for large-scale fortification with folic acid in low-income and lower-middle-income countries (LILMICs) and identify current research related to examining the technological feasibility of fortifying alternative foods with folic acid. (2) Methods: We identified 81 LILMICs, defined by the World Bank’s (WB) 2018 income classifications. To identify dietary intake or apparent consumption, we reviewed WB’s Microdata Library and Global Health Data Exchange for national surveys from 1997–2018. We reviewed survey reports for dietary intake or apparent consumption data and analyzed survey datasets for population coverage of foods. We defined alternative food vehicles as those that may cover/be consumed by ≥30% of the population or households; cereal grains (maize and wheat flours and rice) were included as an alternative food vehicle if a country did not have existing mandatory fortification legislation. To identify current research on fortification with folic acid in foods other than cereal grains, we conducted a systematic review of published literature and unpublished theses, and screened for foods or food products. (3) Results: We extracted or analyzed data from 18 national surveys and countries. The alternative foods most represented in the surveys were oil (n = 16), sugar (n = 16), and salt (n = 14). The coverage of oil ranged from 33.2 to 95.7%, sugar from 32.2 to 98.4%, and salt from 49.8 to 99.9%. We found 34 eligible studies describing research on alternative foods. The most studied alternative foods for fortification with folic acid were dairy products (n = 10), salt (n = 6), and various fruit juices (n = 5). (4) Conclusions: Because of their high coverage, oil, sugar, and salt emerge as potential alternative foods for large-scale fortification with folic acid. However, except for salt, there are limited or no studies examining the technological feasibility of fortifying these foods with folic acid. Full article

To reveal the characteristics and environmental indications for the combination of the grain size and magnetic susceptibility of coastal sediments, we provided a necessary basis for further study on their genetic mechanisms. Based on the data of grain size and magnetic susceptibility of the 36.10 m long core 07SR01 sediments in the Xiyang tidal channel of western South Yellow Sea, we analyzed their variations and correlations and further revealed their environmental indications and corresponding regional sedimentary evolution via the combination of the aforementioned analysis results, the reinterpretation results of the sedimentary sequence and the age of core 07SR01 and shallow seismic profiles, and the findings of climate and glacial–eustatic cycles during Late Quaternary. The three stages of the sedimentary evolution of the Xiyang tidal channel between marine isotope stage (MIS) 7 and MIS 5 were summarized as follows: First is the stage of marginal bank and riverbed developments in the tidal estuary under a relatively high sea level and strong hydrodynamic conditions during MIS 7 (core section: 36.10–26.65 m). The sediments deposited in this stage were mainly affected by the paleo-Changjiang River and characterized by a coarse grain size (mean: 4.02 Φ) and relatively high magnetic susceptibilities (mean: 27.06 × 10⁻⁸ m³·kg⁻¹), with small fluctuations which were strongly and positively correlated with the sand component. Second is the stage dominated by fluviolacustrine and littoral environments with the weak hydrodynamics during MIS 6–5, in which the climate changed from cold and dry to warm and humid as the sea level rose after a drop (core section: 26.65–15.77 m). The sediments deposited in this stage were characterized by a fine grain size (mean: 5.27 Φ) and low magnetic susceptibilities with minor variations (mean: 10.83 × 10⁻⁸ m³·kg⁻¹) which were weakly and positively correlated with the coarse silt component. Third is the stage of delta front in the tidal estuary with a relatively high sea level and strong hydrodynamics during MIS 5 (core section: 15.77–0 m). The sediments deposited in this stage were strongly influenced by the paleo-Yellow River and characterized by a relatively coarse grain size (mean: 4.86 Φ), and high magnetic susceptibilities (mean: 37.15 × 10⁻⁸ m³·kg⁻¹) with large fluctuations which were weakly and positively correlated with the sand and coarse silt components. Full article

Soil salinization is a global issue confronting humanity, imposing significant constraints on agricultural production in the irrigated regions along the southern bank of the Yellow River. This, in turn, leads to the degradation of the ecological environment and inadequate grain yields. Hence, it is essential to explore the magnitude and spatial patterns of soil salinization to promote efficient and sustainable agricultural development. This study carried out a two-year surface soil sampling experiment encompassing the periods before spring irrigation and the budding, flowering, and maturity stages of sunflower fields in the irrigated area along the southern bank of the Yellow River. It employed deep learning in conjunction with multispectral remote sensing conducted by UAV to estimate soil salinity levels in the sunflower fields. Following the identification of sensitive spectral variables through correlation analysis, we proceeded to model and compare the accuracy and stability of various models, including the deep learning Transformer model, traditional machine learning BP neural network (BPNN), random forest model (RF), and partial least squares regression model (PLSR). The findings indicate that the precision of soil salinity content (SSC) retrieval in saline–alkali land can be significantly enhanced by incorporating the RE band of UAV data. Four SSC inversion models were developed using the most suitable spectral variables, resulting in precise soil salinity inversion. The model order based on accuracy and stability was Transformer > BPNN > RF > PLSR. Notably, the Transformer model achieved a prediction accuracy exceeding 0.8 for both the training and test datasets, as indicated by R² values. The precision order of the soil salinity inversion model in each period is as follows: before spring irrigation > budding period > maturity period > flowering stages. Additionally, the accuracy is higher in the bare soil stage compared to the crop cover stage. The Transformer model exhibited RMSE and R² values of 2.41 g kg⁻¹ and 0.84 on the test datasets, with the salt inversion results aligning closely with field-measured data. The results showed that the Transformer deep learning model integrated with RE band data significantly enhances the precision and efficiency of soil salinity inversion within the irrigated regions along the south bank of the Yellow River. Full article