Search Results (6,557)

Non-destructive diameter estimation of cabbage heads and yield prediction employing Unmanned Aerial Vehicle (UAV) imagery are superior to conventional approaches, which are labor intensive and time consuming. This approach assesses spatial variability across the field, effective allocation of resources, and supports variable application rates of fertilizer and supply chain management. Here, individual cabbage head diameters were estimated using deep learning-based pose estimation models (YOLOv8s-pose and YOLOv11s-pose) using high spatial resolution RGB images acquired from UAV 6 m during the cabbage-growing season in 2024. With a mean relative error (MRE) of 4.6% and a high mean average precision (mAP) 98.5% at 0.5, YOLOv11s-pose emerged as the best-performing model, verifying its accuracy for pragmatic agricultural use. The approximated diameter was then combined with climatic variables (temperature and rainfall) and canopy reflectance indices (normalized difference vegetation index (NDVI), normalized difference red edge index (NDRE), and green chlorophyll index (CIg)) that were extracted from the multispectral images with 6 m resolution and fed into AI models to develop individual cabbage head fresh weight. Among the machine learning models (MLMs) tested, CatBoost achieved the lowest Mean Squared Error (MSE = 0.025 kg/cabbage), highest R² (0.89), and outperformed other models based on the Diebold–Mariano statistical test (p < 0.05). This finding suggests that an integrated AI-powered framework enhances non-invasive and precise yield estimation in cabbage farming. Full article

Data preprocessing is usually necessary before running most machine learning classifiers. This work compares three different preprocessing techniques, minimal preprocessing, Principal Components Analysis (PCA), and Linear Discriminant Analysis (LDA). The efficiency of these three preprocessing techniques is measured using the Support Vector Machine (SVM) classifier. Efficiency is measured in terms of statistical metrics such as accuracy, precision, recall, the F-1 measure, and AUROC. The preprocessing times and the classifier run times are also compared using the three differently preprocessed datasets. Finally, a comparison of performance timings on CPUs vs. GPUs with and without the MapReduce environment is performed. Two newly created Zeek Connection Log datasets, collected using the Security Onion 2 network security monitor and labeled using the MITRE ATT&CK framework, UWF-ZeekData22 and UWF-ZeekDataFall22, are used for this work. Results from this work show that binomial LDA, on average, performs the best in terms of statistical measures as well as timings using GPUs or MapReduce GPUs. Full article

One of the essential parts of contemporary cryptographic systems is s-boxes (Substitution Boxes), which give encryption algorithms more complexity and resilience due to their nonlinear structure. In this study, we propose CSBA (Cuckoo Search–Bees Algorithm), a hybrid evolutionary method that combines the strengths of Cuckoo Search and Bees algorithms, to generate s-box structures with strong cryptographic properties. The initial population is generated with a high-diversity four-dimensional Memristive Lu chaotic map, taking advantage of the random yet deterministic nature of chaotic systems. This proposed method was designed with inspiration from biological systems. It was developed based on the foraging strategies of bees and the reproductive strategies of cuckoos. This nature-inspired structure enables an efficient scanning of the solution space. The resultant s-boxes’ fitness was assessed using the nonlinearity value. These s-boxes were then optimized using the hybrid CSBA algorithm suggested in this paper as well as the Bees algorithm. The performance of the proposed approaches was measured using SAC, nonlinearity, BIC-SAC, BIC-NL, maximum difference distribution, and linear uniformity (LU) metrics. Compared to other studies in the literature that used metaheuristic algorithms to generate s-boxes, the proposed approach demonstrates good performance. In particular, the average value of 109.75 obtained for the nonlinearity metric demonstrates high success. Therefore, this study demonstrates that robust and reliable s-boxes can be generated for symmetric encryption algorithms using the developed metaheuristic algorithms. Full article

Multimodal remote sensing object detection has gained increasing attention due to its ability to leverage complementary information from different sensing modalities, particularly visible (RGB) and thermal infrared (TIR) imagery. However, existing methods typically depend on deep, computationally intensive backbones and complex fusion strategies, limiting their suitability for real-time applications. To address these challenges, we propose a lightweight and efficient detection framework named RGB-TIR Multimodal Fusion Network (RTMF-Net), which introduces innovations in both the backbone architecture and fusion mechanism. Specifically, RTMF-Net adopts a dual-stream structure with modality-specific enhancement modules tailored for the characteristics of RGB and TIR data. The visible-light branch integrates a Convolutional Enhancement Fusion Block (CEFBlock) to improve multi-scale semantic representation with low computational overhead, while the thermal branch employs a Dual-Laplacian Enhancement Block (DLEBlock) to enhance frequency-domain structural features and weak texture cues. To further improve cross-modal feature interaction, a Weighted Denoising Fusion Module is designed, incorporating an Enhanced Fusion Attention (EFA) attention mechanism that adaptively suppresses redundant information and emphasizes salient object regions. Additionally, a Shape-Aware Intersection over Union (SA-IoU) loss function is proposed to improve localization robustness by introducing an aspect ratio penalty into the traditional IoU metric. Extensive experiments conducted on the ODinMJ and LLVIP multimodal datasets demonstrate that RTMF-Net achieves competitive performance, with mean Average Precision (mAP) scores of 98.7% and 95.7%, respectively, while maintaining a lightweight structure of only 4.3M parameters and 11.6 GFLOPs. These results confirm the effectiveness of RTMF-Net in achieving a favorable balance between accuracy and efficiency, making it well-suited for real-time remote sensing applications. Full article

Background: Gamification may enhance engagement and higher-order learning in health-care profession education, but evidence from undergraduate nursing programs—particularly in the Middle East—is limited. We evaluated whether integrating structured gamified activities into an anatomy and physiology course improves class engagement and knowledge-based critical thinking. Methods: In this pragmatic, nonrandomized, section-allocated quasi-experimental study at a single Saudi institution, 121 first-year female nursing students were assigned by existing cohorts to traditional instruction (control; n = 61) or instruction enhanced with gamified elements (intervention; n = 60) groups. The intervention (introduced mid-semester) comprised time-limited competitive quizzing with immediate feedback and aligned puzzle tasks. Outcomes were measured at baseline, mid-semester, and end-semester using a four-item Class Engagement Rubric (CER; scale 1–5) and a 40-item high-cognitive multiple-choice (MCQ) assessment mapped to course objectives. Analyses used paired and independent t-tests with effect sizes and 95% confidence intervals. Results: No attrition occurred. From baseline to end-semester, the intervention group had a mean CER increase of 0.59 points (95% CI, 0.42 to 0.76; p < 0.001)—approximately a 15% relative gain—and a mean MCQ increase of 0.30 points (95% CI, 0.18 to 0.42; p < 0.001), an ~8% relative gain. The control group showed no material change over the same interval. Between-group differences in change favored the intervention across CER items and for the MCQ outcome. Semester grade-point average did not differ significantly between groups (p = 0.055). Conclusions: Embedding a brief, structured gamification package within an undergraduate nursing anatomy and physiology course was associated with measurable improvements in classroom engagement and modest gains in knowledge-based critical thinking, with no detectable effect on overall semester GPA. Given the nonrandomized, single-site design, causal inference is limited. Multi-site randomized trials using validated critical-thinking instruments are warranted to confirm effectiveness and define dose, durability, and generalizability. Full article

This study proposes a lightweight real-time fishing vessel tracking and size measurement model based on a UAV. In view of the problems faced by the current fishing port management department, such as low efficiency of fishing vessel size measurement methods and difficulty in updating the size information of large quantities of fishing vessels in time, this paper proposes a lightweight real-time fishing vessel tracking and size measurement model based on a UAV. (YOLO-LFVM). The model incorporates lightweight modules, such as MobileNetV3, AKConv, and C2f, and utilizes Python scripts in conjunction with OpenCV to measure vessel size in pixels. The findings indicate that, compared to the original model, the YOLO-LFVM model’s accuracy rate, recall rate, and mAP@0.5 decrease by only 0.7%, 0.2%, and 0.3%, respectively, while mAP@0.95 increases by 1.7%. Additionally, the model’s parameters decrease by 65%, and GFLOPs decrease by 69%. When comparing the model’s output with actual vessel data, the average relative error for total length is 2.67%, and for width, it is 3.28%. The research shows that the YOLO-LFVM model is effective in ship identification, ship tracking statistics, and measurement. Through the integration with UAV remote sensing technology, it is conducive to the timely updating of large-scale fishing vessel size information. Finally, the model can assist the daily management and law enforcement of the fishing port management department and can be applied to other equipment with limited computing power to perform target detection and object size measurement tasks. Full article

Background: Chemical burns represent a persistent global health challenge due to their high prevalence, causing lifelong disabilities and socioeconomic burdens. Although research on chemical burns has expanded over the past century, no comprehensive study has mapped the intellectual structure, global collaboration patterns, and thematic evolution of scientific production on chemical burns to determine how research in the area has evolved and the existence of gaps or imbalances that need to be addressed. Objective: The aim was to analyze the scientific production on chemical burns using bibliometric methods, identifying key contributors, evolving themes, and research gaps. Methods: Eligible documents contained the MeSH descriptor and were listed both in PubMed (1946 to 2024) and in the Web of Science Core Collection. The documents were analyzed with Bibliometrix version 5.0 and VOSviewer version 1.6.20. The metrics included were annual productivity, citation networks, co-authorship patterns, and keyword co-occurrence. Results: The analysis included 3943 articles from 757 journals. The annual average was 25.8 articles, with a growth rate of 0.65% from 1946 to 2024. The USA produced the most articles (n = 1547), followed by China (n = 890). The USA also led in international collaboration, working with 26 countries. Harvard University was the leading institution (n = 325) and Burns the leading journal (n = 306), followed by Cornea (n = 132). The most common subject category of the research was surgery (n = 1185 docs) and ophthalmology (n = 984). Reim M. was the most prolific author (n = 35), while Basu S. had the most citations (n = 1159). The main clinical MeSH descriptors were “Eye burns” (n = 1158), “Esophageal stenosis” (n = 683), and “Caustics” (n = 659). Conclusions: The results show slight growth in scientific production on chemical burns. The USA and China are leading research in this field, and the main reported finding was eye burns. Full article

Under night vision conditions, visible-spectrum images often fail to capture background details. Conventional visible and infrared fusion methods generally overlay thermal signatures without preserving latent features in low-visibility regions. This paper proposes a novel deep learning-based fusion algorithm to enhance visual perception in night driving scenarios. Firstly, a local adaptive enhancement algorithm corrects underexposed and overexposed regions in visible images, thereby preventing oversaturation during brightness adjustment. Secondly, ResNet152 extracts hierarchical feature maps from enhanced visible and infrared inputs. Max pooling and average pooling operations preserve critical features and distinct information across these feature maps. Finally, Linear Discriminant Analysis (LDA) reduces dimensionality and decorrelates features. We reconstruct the fused image by the weighted integration of the source images. The experimental results on benchmark datasets show that our approach outperforms state-of-the-art methods in both objective metrics and subjective visual assessments. Full article

During the last decades, there have been increasing concerns in public health debates regarding the production and consumption of red meat, considering connections between the occurrence of nutrition transition and an increase in the prevalence of chronic noncommunicable diseases. The consumption of red meat has been linked to adverse health outcomes; however, current evidence reveals controversies regarding the intake of diverse red meats. In addition, barriers to meat consumption include sanitary legislation linked to foodborne diseases connected to livestock, whilst governments of diverse countries provide incentives for its production and export worldwide. Thus, the objective of the present study was to investigate the evolution in the global trade of processed and unprocessed red meat from 1986 to 2023, using network analysis. Data on the trade of red meat between pairs of 216 countries were obtained from the Food and Agriculture Organization Database (FAOSTAT). The dataset, comprising the mean annual volume of processed and unprocessed red meat exchanged from reporting countries (origin) to partner countries (destination), was used to map global trade networks of red meats and identify global trends in red meat consumption according to country income level. The results indicate substantial intensification in the global trade of processed (0.202 in 1986 to 0.453 kg per capita in 2023) and unprocessed red meat (1.415 in 1986 to 3.315 Kg per capita in 2023). The volume of trade of unprocessed red meat remains greater than the volume processed red meat; yet, the findings indicate a threefold increase in the average weighted degree of processed red meat trade (0.002 to 0.006) from 1986 until 2023, whilst unprocessed red meat showed a twofold increase (0.009 to 0.019). The results raise public health concerns regarding the long-term consequences of consuming processed foods with high sodium and fat content. Additionally, the global trade of red meat showed fluctuations in periods of major foodborne outbreaks related to meat consumption, particularly during the 1990s. The findings of the study highlight strategies at the national level to advance food system transformations towards improvements in public health, nutrition, and sustainability. Full article

Monitoring the concentration of heavy metals in urban soils is of a paramount importance for several reasons. These inorganic pollutants can pose a significant health risk to living organisms, as they are toxic even at low concentrations and can be present in the soil for a long period of time. This study assesses the spatial distribution, concentration levels, and potential anthropogenic and natural sources of eight typical heavy metals (As, Cd, Co, Cr, Cu, Ni, Pb and Zn) occurring in urban surface soils across Debrecen, Hungary. A total of 295 topsoil samples were collected; heavy metal concentrations were determined by energy-dispersive X-ray fluorescence (EDXRF) spectrometry. The results were interpreted using descriptive statistics, correlation analysis, hierarchical clustering, factor analysis, ordinary kriging interpolation, and spatial-discriminant analysis. The dual origin of the metal contaminants was revealed: As, Co, Pb, and Zn showed strong anthropogenic signatures associated with traffic, urban waste, and construction materials, whereas Cr and Ni were associated with natural geogenic sources. Cd reflected both lithogenic and point-source urban pollution. The current evaluation incorporated Hungarian and Dutch regulatory benchmarks to identify exceedances of environmental quality thresholds. It was found that only Cd and Cr exceeded the Hungarian target values, on average. Linear discriminant analysis based on pollution maps highlighted contamination hotspots around traffic corridors and newly industrialized zones. The importance of high-resolution soil monitoring in the rapidly urbanizing city is highlighted. Given its anticipated industrial and transportation developments, accumulations of heavy metals are probably going to be further exacerbated; therefore, the results provide a critical baseline for future environmental assessments and long-term monitoring. Full article

In real-world dairy farming environments, object recognition models often suffer from missed or false detections due to complex backgrounds and cow occlusions. In response to these issues, this paper proposes FSCA-YOLO, a multi-object cow behavior recognition model based on an improved YOLOv11 framework. First, the FEM-SCAM module is introduced along with the CoordAtt mechanism to enable the model to better focus on effective behavioral features of cows while suppressing irrelevant background information. Second, a small object detection head is added to enhance the model’s ability to recognize cow behaviors occurring at the distant regions of the camera’s field of view. Finally, the original loss function is replaced with the SIoU loss function to improve recognition accuracy and accelerate model convergence. Experimental results show that compared with mainstream object detection models, the improved YOLOv11 in this section demonstrates superior performance in terms of precision, recall, and mean average precision (mAP), achieving 95.7% precision, 92.1% recall, and 94.5% mAP—an improvement of 1.6%, 1.8%, and 2.1%, respectively, over the baseline YOLOv11 model. FSCA-YOLO can accurately extract cow features in real farming environments, providing a reliable vision-based solution for cow behavior recognition. To support specific behavior recognition and in-region counting needs in multi-object cow behavior recognition and tracking systems, OpenCV is integrated with the recognition model, enabling users to meet the diverse behavior identification requirements in groups of cows and improving the model’s adaptability and practical utility. Full article

Accurate assessment of cropland maps is crucial for ensuring food security, effective agricultural management, and environmental monitoring. With the widespread application of high-resolution (≤10 m) remote sensing imagery and the advancement of machine learning techniques, numerous high-resolution cropland maps have been developed. However, comprehensive evaluations of their accuracy remain limited. We utilized 163,861 validation samples and national land survey statistical data to conduct a multi-scale comparison of the accuracy of seven cropland maps (one 1 m and six 10 m maps) in China. Additionally, five synergistic strategies were employed to generate more accurate fused cropland maps. Validation results showed that the overall accuracy (OA) of the seven maps ranged from 0.79 to 0.91, with ESA-WorldCover (ESA-WC) exhibiting the highest OA, followed by AI Earth China land cover classification dataset (AIEC), ESRI Land Cover (ESRI-LC), and Cropland Use Intensity in China (China-CUI), while Sino-LC1 showed the lowest performance. Spatially, ESA-WC achieved the highest accuracy in nearly 60% of provinces, followed by AIEC and ESRI-LC, each accounting for approximately 20%. AIEC performed best in western provinces, whereas ESRI-LC dominated in the middle and lower reaches of the Yangtze River. Area consistency assessments revealed that, on average, the seven maps overestimated cropland areas by 20% compared to statistical data. Among these, ESA-WC showed the highest proportion of provinces with relative errors within ±20%, but this proportion was only 50%. Moreover, the OA of the fused maps exceeded 0.92, with county-level R² values compared to statistical data reaching 0.98, significantly improving the reliability of cropland products in over 60% of provincial administrative regions. Based on these results, effective synergistic strategies for high-resolution cropland mapping are proposed. Full article

Background: Climate change threatens global food security, highlighting the need for adaptive traits in livestock to ensure sustainable production. Rabbits, known for their unique adaptability, require the preservation of genetic diversity to maintain resilience. The decline in genetic specificity among indigenous breeds underscores the urgency of conservation efforts to protect these critical resources. Objectives: This study investigates the genetic structure and diversity of indigenous rabbit populations, emphasizing genetic mapping as essential for sustaining adaptability. The findings aim to guide breeding programs that enhance biodiversity and support agricultural resilience. Materials and Methods: This study analyzed both native and exotic rabbit breeds. Native breeds included Black Baladi (BB), White Baladi (WB), Red Baladi (RB), and Jabali (JAB), while exotic breeds included New Zealand White (NZW), American Rex (AR), and Chinchilla (CH). Fourteen microsatellite loci were genotyped in 526 rabbits across all breeds. Results: A total of 467 alleles were identified, with an overall mean of 5.03. The expected heterozygote frequencies were medium to high. Polymorphism was high in BB, JAB, and NZW, and medium in WB, RB, AR, and CH. F_IS and F_IT values (−0.044 and 0.156) suggested possible non-intensive inbreeding. F_ST (0.220) showed breed differentiation and high within-breed variation. The gene flow averaged 1.872, indicating interbreed gene exchange. Neutrality and phylogenetic analyses revealed genetic reshaping; BB, WB, RB, AR, CH, and NZW showed overlap, while JAB retained high specificity. Conclusions: Urgent conservation strategies are essential to preserve native rabbit genetic diversity and unique traits, which are vital for sustaining biodiversity and livestock resilience globally. Full article

Diabetic Retinopathy (DR) is a leading cause of vision impairment globally, underscoring the need for accurate and early diagnosis to prevent disease progression. Although fundus imaging serves as a cornerstone of Computer-Aided Diagnosis (CAD) systems, several challenges persist, including lesion scale variability, blurry morphological patterns, inter-class imbalance, limited labeled datasets, and computational inefficiencies. To address these issues, this study proposes an end-to-end diagnostic framework that integrates an enhanced preprocessing pipeline with a novel deep learning architecture, Hierarchical-Inception-Residual-Dense Network (HIRD-Net). The preprocessing stage combines Contrast Limited Adaptive Histogram Equalization (CLAHE) with Dilated Difference of Gaussian (D-DoG) filtering to improve image contrast and highlight fine-grained retinal structures. HIRD-Net features a hierarchical feature fusion stem alongside multiscale, multilevel inception-residual-dense blocks for robust representation learning. The Squeeze-and-Excitation Channel Attention (SECA) is introduced before each Global Average Pooling (GAP) layer to refine the Feature Maps (FMs). It further incorporates four GAP layers for multi-scale semantic aggregation, employs the Hard-Swish activation to enhance gradient flow, and utilizes the Focal Loss function to mitigate class imbalance issues. Experimental results on the IDRiD-APTOS2019, DDR, and EyePACS datasets demonstrate that the proposed framework achieves 93.46%, 82.45% and 79.94% overall classification accuracy using only 4.8 million parameters, highlighting its strong generalization capability and computational efficiency. Furthermore, to ensure transparent predictions, an Explainable AI (XAI) approach known as Gradient-weighted Class Activation Mapping (Grad-CAM) is employed to visualize HIRD-Net’s decision-making process. Full article

Labor shortages and uneven terrain in orchards present significant challenges to autonomous navigation. This study proposes a navigation system that integrates Light Detection and Ranging (LiDAR) and Inertial Measurement Unit (IMU) data to enhance localization accuracy and map stability through Simultaneous Localization and Mapping (SLAM). To minimize distortions in LiDAR scans caused by ground irregularities, real-time tilt correction was implemented based on IMU feedback. Furthermore, the path planning module was improved by modifying the Rapidly-Exploring Random Tree (RRT) algorithm. The enhanced RRT generated smoother and more efficient trajectories with quantifiable improvements: the average shortest path length was 2.26 m, compared to 2.59 m with conventional RRT and 2.71 m with A* algorithm. Tracking performance also improved, achieving a root mean square error of 0.890 m and a maximum lateral deviation of 0.423 m. In addition, yaw stability was strengthened, as heading fluctuations decreased by approximately 7% relative to the standard RRT. Field results validated the robustness and adaptability of the proposed system under real-world agricultural conditions. These findings highlight the potential of LiDAR–IMU sensor fusion and optimized path planning to enable scalable and reliable autonomous navigation for precision agriculture. Full article