Search Results (1,843)

Construction monitoring of cutter suction dredgers (CSD) is of great significance in ensuring dredging efficiency. However, existing models have not taken into account the physical constraints in the physical system of a CSD, which limits further improvements in prediction accuracy. To this end, this paper proposes a physics-assisted deep learning model for improved construction performance monitoring of CSD. The data-driven Loss_u and the physics-driven Loss_r are combined to form an improved physics-assisted loss function (PALF). And then, a physics-assisted deep learning (PADL) model incorporating PALF is developed to predict the construction productivity. In the case application, evaluation across three deep learning models confirms the feasibility and effectiveness of PALF for productivity prediction. The results show that the PALF-based PADL model achieves markedly improved prediction accuracy, reducing the mean absolute error by 20.33–54.33%. Across six training-set sizes (1000–11,000 samples), the improvement is larger in small-data scenarios, highlighting PADL’s strong low-sample robustness. The proposed model can effectively complement physical sensors in monitoring construction parameters and provide reliable decision support for assessing the operational state of CSDs. Full article

Hyperspectral imagery (HSI), as a core data carrier in remote sensing, plays a crucial role in many fields. Still, it also faces numerous challenges, including the curse of dimensionality, noise interference, and small samples. These problems severely affect the generalization ability and classification accuracy of traditional machine learning and deep learning algorithms. Existing solutions suffer from bottlenecks such as unknown cost matrices and excessive computational overhead. And ensemble learning fails to fully exploit the deep semantic features and feature importance relationships of high-dimensional data. To address these issues, this paper proposes a dual ensemble classification framework (DRF-FI) based on feature importance analysis and a deep random forest. This method integrates feature selection and two-layer ensemble learning. First, it identifies discriminative spectral bands through feature importance quantification. Then, it constructs a balanced training subset through random oversampling. Finally, it integrates four different ensemble strategies. Experimental results on three benchmark hyperspectral datasets demonstrate that DRF-FI exhibits outstanding performance across multiple datasets, particularly excelling in handling highly imbalanced data. Compared to traditional random forests, the proposed method achieves stable improvements in both overall accuracy (OA) and average accuracy (AA). On specific datasets, OA and AA were enhanced by up to 0.84% and 1.24%, respectively. This provides an effective solution to the class imbalance problem in hyperspectral images. Full article

Deep learning (DL), a hierarchical feature extraction method, has garnered increasing attention in the remote sensing community. Recently, self-supervised learning (SSL) methods in DL have gained wide recognition due to their ability to mitigate the dependence on both the quantity and quality of samples. This advantage is particularly significant when dealing with limited labeled samples in hyperspectral images (HSIs). However, conventional SSL methods face two main challenges. They struggle to construct self-supervised signals based on the unique characteristics of HSI. Moreover, they fail to design network optimization strategies that leverage the intrinsic manifold geometry within HSI. To tackle these issues, we propose a novel self-supervised learning method termed Manifold Geometry-Leveraged Self-supervised Learning (MSSL) for HSI classification. The approach employs a two-stage training strategy. In the initial pre-training stage, it develops self-supervised signals that exploit spatial homogeneity and spectral coherence properties of HSI. Furthermore, it introduces a manifold geometry-guided loss function that enhances feature discrimination by increasing intra-class compactness and inter-class separation. The second stage is a fine-tuning phase utilizing a small set of labeled samples. This stage optimizes the pre-trained model, enabling effective feature extraction from hyperspectral data for classification tasks. Experiments conducted on real-world HSI datasets demonstrate that MSSL achieves superior classification performance compared to several relevant state-of-the-art methods. Full article

With the rapid development of remote sensing technology and deep learning, the port ship detection based on a single-stage algorithm has achieved remarkable results in optical imagery. However, most of the existing methods are designed and verified in specific scenes, such as fixed viewing angle, uniform background, or open sea, which makes it difficult to deal with the problem of ship detection in complex environments, such as cloud occlusion, wave fluctuation, complex buildings in the harbor, and multi-ship aggregation. To this end, ES-YOLO framework is proposed to solve the limitations of ship detection. A novel edge perception channel, Spatial Attention Mechanism (EACSA), is proposed to enhance the extraction of edge information and improve the ability to capture feature details. A lightweight spatial–channel decoupled down-sampling module (LSCD) is designed to replace the down-sampling structure of the original network and reduce the complexity of the down-sampling stage. A new hierarchical scale structure is designed to balance the detection effect of different scale differences. In this paper, a remote sensing ship dataset, TJShip, is constructed based on Gaofen-2 images, which covers multi-scale targets from small fishing boats to large cargo ships. The TJShip dataset was adopted as the data source, and the ES-YOLO model was employed to conduct ablation and comparison experiments. The results show that the introduction of EACSA attention mechanism, LSCD, and multi-scale structure improves the mAP of ship detection by 0.83%, 0.54%, and 1.06%, respectively, compared with the baseline model, also performing well in precision, recall and F1. Compared with Faster R-CNN, RetinaNet, YOLOv5, YOLOv7, and YOLOv8 methods, the results show that the ES-YOLO model improves the mAP by 46.87%, 8.14%, 1.85%, 1.75%, and 0.86%, respectively, under the same experimental conditions, which provides research ideas for ship detection. Full article

Citrus orchards are especially vulnerable owing to low inter-row vegetation cover, and frequent tillage. Here, we combine controlled field experiments with proximal remote sensing–derived geomorphometric variables and machine learning (ML) to identify key factors of erosion in a Mediterranean climate citrus plantation located close to Seville and the National Park of Doñana (Southern Spain) on Gleyic Regosols (clayic, arenic). We conducted rainfall simulations with 30 s sampling, measured infiltration (mini-disc infiltrometer), saturated hydraulic conductivity (Kfs; Guelph permeameter), compaction (penetrologger), and soil respiration (gas analyzer) at multiple points, and derived high resolution morphometric indices from proximal sensing (UAV-LiDAR). Linear models and Random Forests were trained to explain three responses: soil loss, sediment concentration (SC), and runoff. Results show that soil loss is most strongly associated with maximum compaction and Kfs (multiple regression: R² = 0.68; adjusted R² = 0.52; p = 0.063), while SC increases with surface compaction and exhibits weak relationships with topographic metrics. Runoff decreases with average infiltration, which is related to compaction (β = −4.83 ± 2.38; R² = 0.34; p = 0.077). Diagnostic checks indicate centered residuals with mild heteroscedasticity and a few high leverage observations. Random Forests captured part of the variance for soil loss (≈29%) but performed poorly for runoff, consistent with limited sample size and modest nonlinear signal. Morphometric analysis revealed gentle relief but pronounced convergent–divergent patterns that modulate hydrological connectivity. There were strong differences in the experiments conducted close to the trees and in the tractor trails. We conclude that compaction and near surface hydraulic properties are the most influential and measurable controls of erosion at plot scale and the UAV-LiDAR could not give us extra-insights. We highlight that integrating standardized field protocols with proximal morphometrics and ML can be the best method to prioritize a small set of explanatory variables, helping to reduce experimental effort while maintaining explanatory power. Full article

Background/Objectives: Traditional dietary monitoring methods such as 24 h recalls rely on self-report, leading to recall bias and underreporting. Similarly, dietary control approaches, including portion control and calorie restriction, depend on user accuracy and consistency. Augmented reality (AR) offers a promising alternative for improving dietary monitoring and control by enhancing engagement, feedback accuracy, and user learning. This systematic review aimed to examine how AR technologies are implemented to support dietary monitoring and control and to evaluate their usability and effectiveness among adults. Methods: A systematic search of PubMed, CINAHL, and Embase identified studies published between 2000 and 2025 that evaluated augmented reality for dietary monitoring and control among adults. Eligible studies included peer-reviewed and gray literature in English. Data extraction focused on study design, AR system type, usability, and effectiveness outcomes. Risk of bias was assessed using the Cochrane RoB 2 tool for randomized controlled trials and ROBINS-I for non-randomized studies. Results: Thirteen studies met inclusion criteria. Since the evidence based was heterogeneous in design, outcomes, and measurement, findings were synthesized qualitatively rather than pooled. Most studies utilized smartphone-based AR systems for portion size estimation, nutrition education, and behavior modification. Usability and satisfaction varied by study: One study found that 80% of participants (N = 15) were satisfied or extremely satisfied with the AR tool. Another reported that 100% of users (N = 26) rated the app easy to use, and a separate study observed a 72.5% agreement rate on ease of use among participants (N = 40). Several studies also examined portion size estimation, with one reporting a 12.2% improvement in estimation accuracy and another showing −6% estimation, though a 12.7% overestimation in energy intake persisted. Additional outcomes related to behavior, dietary knowledge, and physiological or psychological effects were also identified across the review. Common limitations included difficulty aligning markers, overestimation of amorphous foods, and short intervention durations. Despite these promising findings, the existing evidence is limited by small sample sizes, heterogeneity in intervention and device design, short study durations, and variability in usability and accuracy measures. The limitations of this review warrant cautious interpretation of findings. Conclusions: AR technologies show promise for improving dietary monitoring and control by enhancing accuracy, engagement, and behavior change. Future research should focus on longitudinal designs, diverse populations, and integration with multimodal sensors and artificial intelligence. Full article

Microscopic wood images are vital in wood analysis and classification research. However, the high cost of acquiring microscopic images and the limitations of experimental conditions have led to a severe problem of insufficient sample data, which significantly restricts the training performance and generalization ability of deep learning models. This study first used basic image processing techniques to perform preliminary augmentation of the original dataset. The augmented data were then input into five GAN models, BGAN, DCGAN, WGAN-GP, LSGAN, and StyleGAN2, for training. The quality and model performance of the generated images were assessed by analyzing the degree of fidelity of cellular structure (e.g., earlywood, latewood, and wood rays), image clarity, and diversity of the images for each model-generated image, as well as by using KID, IS, and SSIM. The results showed that images generated by BGAN and WGAN-GP exhibited high quality, with lower KID values and higher IS values, and the generated images were visually close to real images. In contrast, the DCGAN, LSGAN, and StyleGAN2 models experienced mode collapse during training, resulting in lower image clarity and diversity compared to the other models. Through a comparative analysis of different GAN models, this study demonstrates the feasibility and effectiveness of Generative Adversarial Networks in the domain of small-sample image data augmentation, providing an important reference for further research in the field of wood identification. Full article

The re-identification of endangered felines is critical for species conservation and biodiversity assessment. This paper proposes the Pose-Guided Network with the Adaptive L2 Regularization (PGNet-AL2) framework to overcome key challenges in wild feline re-identification, such as extensive pose variations, small sample sizes, and inconsistent image quality. This framework employs a dual-branch architecture for multi-level feature extraction and incorporates an adaptive L2 regularization mechanism to optimize parameter learning, effectively mitigating overfitting in small-sample scenarios. Applying the proposed method to the Amur Tiger Re-identification in the Wild (ATRW) dataset, we achieve a mean Average Precision (mAP) of 91.3% in single-camera settings, outperforming the baseline PPbM-b (Pose Part-based Model) by 18.5 percentage points. To further evaluate its generalization, we apply it to a more challenging task, snow leopard re-identification, using a dataset of 388 infrared videos obtained from the Wildlife Conservation Society (WCS). Despite the poor quality of infrared videos, our method achieves a mAP of 94.5%. The consistent high performance on both the ATRW and snow leopard datasets collectively demonstrates the method’s strong generalization capability and practical utility. Full article

Background/Objectives: Quantitative computed tomography (CT) metrics are widely used to assess pulmonary involvement and to predict short-term severity in coronavirus disease 2019 (COVID-19). However, it remains unclear whether baseline artificial intelligence (AI)-based quantitative high-resolution computed tomography (HRCT) metrics of pneumonia burden provide incremental prognostic value for in-hospital composite adverse outcomes beyond routine clinical factors, or whether these imaging-derived markers carry any exploratory signal for long-term severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) reinfection among hospitalized patients. Most existing imaging studies have focused on diagnosis and acute-phase prognosis, leaving a specific knowledge gap regarding AI-based quantitative HRCT correlates of early deterioration and subsequent reinfection in this population. To evaluate whether combining deep learning-derived, quantitative, HRCT features and clinical factors improve prediction of in-hospital composite adverse events and to explore their association with long-term reinfection in patients with COVID-19 pneumonia. Methods: In this single-center retrospective study, we analyzed 236 reverse-transcription polymerase chain reaction (RT-PCR)-confirmed COVID-19 patients who underwent baseline HRCT. Median follow-up durations were 7.65 days for in-hospital outcomes and 611 days for long-term outcomes. A pre-trained, adaptive, artificial-intelligence-based, prototype model (Siemens Healthineers) was used for pneumonia analysis. Inflammatory lung lesions were automatically segmented, and multiple quantitative metrics were extracted, including opacity score, volume and percentage of opacities and high-attenuation opacities, and mean Hounsfield units (HU) of the total lung and opacity. Patients were stratified based on receiver operating characteristic (ROC)-derived optimal thresholds, and multivariable Cox regression was used to identify predictors of the composite adverse outcome (intensive care unit [ICU] admission or all-cause death) and SARS-CoV-2 reinfection, defined as a second RT-PCR-confirmed episode of COVID-19 occurring ≥90 days after initial infection. Results: The composite adverse outcome occurred in 38 of 236 patients (16.1%). Higher AI-derived opacity burden was significantly associated with poorer outcomes; for example, opacity score cut-off of 5.5 yielded an area under the ROC curve (AUC) of 0.71 (95% confidence interval [CI] 0.62–0.79), and similar performance was observed for the volume and percentage of opacities and high-attenuation opacities (AUCs up to 0.71; all p < 0.05). After adjustment for age and comorbidities, selected HRCT metrics—including opacity score, percentage of opacities, and mean HU of the total lung (cut-off −662.38 HU; AUC 0.64, 95% CI 0.54–0.74)—remained independently associated with adverse events. Individual predictors demonstrated modest discriminatory ability, with C-indices of 0.59 for age, 0.57 for chronic obstructive pulmonary disease (COPD), 0.62 for opacity score, 0.63 for percentage of opacities, and 0.63 for mean total-lung HU, whereas a combined model integrating clinical and imaging variables improved prediction performance (C-index = 0.68, 95% CI: 0.57–0.80). During long-term follow-up, RT-PCR–confirmed reinfection occurred in 18 of 193 patients (9.3%). Higher baseline CT-derived metrics—particularly opacity score and both volume and percentage of high-attenuation opacities (percentage cut-off = 4.94%, AUC 0.69, 95% CI 0.60–0.79)—showed exploratory associations with SARS-CoV-2 reinfection. However, this analysis was constrained by the very small number of events (n = 18) and wide confidence intervals, indicating substantial statistical uncertainty. In this context, individual predictors again showed only modest C-indices (e.g., 0.62 for procalcitonin [PCT], 0.66 for opacity score, 0.66 for the volume and 0.64 for the percentage of high-attenuation opacities), whereas the combined model achieved an apparent C-index of 0.73 (95% CI 0.64–0.83), suggesting moderate discrimination in this underpowered exploratory reinfection sample that requires confirmation in external cohorts. Conclusions: Fully automated, deep learning-derived, quantitative HRCT parameters provide useful prognostic information for early in-hospital deterioration beyond routine clinical factors and offer preliminary, hypothesis-generating insights into long-term reinfection risk. The reinfection-related findings, however, require external validation and should be interpreted with caution given the small number of events and limited precision. In both settings, combining AI-based imaging and clinical variables yields better risk stratification than either modality alone. Full article

Accurate indoor floor classification is essential for wireless positioning systems. However, the performance of conventional received signal strength indictor (RSSI)-based fingerprinting approaches is often limited by signal fluctuations and insufficient feature representation. To address these challenges, this paper introduces a heterogeneous graph neural network (GNN) framework that models WiFi signals using two types of nodes: reference points and Media Access Control (MAC) address. The edges between reference points and MAC addresses are weighted by normalized RSSI values, allowing the model to capture signal strength interactions and perform relation-aware message passing. Through this graph-based representation, the model can learn spatial and signal dependencies more effectively than conventional vector-based approaches. The proposed model was extensively evaluated under both benchmark and practical settings. On small-scale datasets, it achieved performance comparable to that of a conventional convolutional neural network trained on large-scale datasets, confirming its effectiveness with limited samples. In addition, the proposed model consistently outperformed other models under noisy conditions, achieving 93.88% accuracy on the widely used UJIIndoorLoc dataset and 97.3% accuracy in real-time experiments conducted at a test site. These values are significantly higher than those achieved using conventional machine learning (ML) baselines, highlighting the ability of the proposed model to handle real-world signal variations. These findings highlight that the heterogeneous GNN effectively captures spatial and signal-level dependencies, offering a robust and scalable solution for accurate indoor floor classification. Overall, this work presents a promising pathway for improving the performance and reliability of future wireless positioning systems. Full article

Financial distress prediction models are widely used to support risk management. However, economic turbulence, such as the COVID-19 pandemic, can disrupt the relationships between financial indicators and distress, thus threatening the stability and accuracy of the models’ predictions. In this study, the stability of bankruptcy prediction models is examined on a large sample of small and medium-sized enterprises (SMEs) in Slovakia. Three periods are distinguished: the pre-pandemic years 2018–2019, the COVID-19 pandemic years 2020–2021, and the post-pandemic recovery years 2022–2023. Two approaches to model construction are compared: separate models are estimated for each period, and a single comprehensive model covering all three periods is constructed with a period-specific indicator among the predictors. Publicly available financial data and machine learning methods are employed, and model performance is evaluated using common classification metrics. Differences in performance are revealed, indicating whether period-specific models provide superior predictive accuracy or whether a universal model can adapt to changing economic conditions. The robustness, stability, predictive power, and practical applicability of both approaches are assessed, and the influence of economic fluctuations on accuracy is demonstrated. The findings provide guidance on selecting modelling strategies across different economic environments and offer recommendations for further developing and implementing predictive models in volatile financial conditions. Full article

Background: Group-based dance interventions (GBDIs) have emerged as a promising approach to rehabilitation for neurological disorders. This pilot study evaluated the feasibility and preliminary effects of a GBDI on motor function, cognition, fatigue, and quality of life in individuals with Relapsing–Remitting Multiple Sclerosis (RRMS). Methods: The intervention consisted of two 60-min ballet sessions per week over 10 weeks, structured as 10 min of warm-up, 40 min of ballet exercises, and 10 min of stretching. Assessments were conducted at baseline (T0) and post-intervention (T1). Concerning motor measures, balance was assessed using the Mini-BESTest; gait performance was evaluated through the 6-min walk test (6MWT), four square step test (FSST), and figure-of-8 walk test (F8WT); upper limb motor functions were assessed using the box and block test (BBT) and 9-hole peg test (9HPT). Regarding cognitive functions, the Rey auditory verbal learning test (RAVLT), symbol digit modalities test (SDMT), and trail making test A and B (TMT-A/B) were administered, while fatigue and quality of life were assessed using the modified fatigue impact scale (MFIS) and the Short Form survey-36 (SF-36), respectively. Results: At T1, participants improved in Mini-BESTest (+17.5%), 6MWT (+7.3%), and BBT dominant hand (+6.9%). Performance also improved on the following cognitive tests: RAVLT Immediate Recall (+5.9%), RAVLT Delayed Recall (+20.3%), SDMT (+47.4%), TMT-A (−21.2%), and (TMT-B −24.5%). Conclusions: The very small sample size (n = 4) and the lack of a control group probably restrict the generalizability of the findings. Consequently, the results obtained by this pilot study should be considered exploratory and hypothesis-generating rather than definitive evidence of a robust benefit. Future studies should confirm these findings by enlarging the intervention cohorts and adopting a randomized controlled design. In this sense, a 10-week GBDI may provide a solid base for a safe and promising dance-based rehabilitation program that could lead to improvements in motor, cognition, and psychosocial spheres in people with RRMS. Full article

The relationship between retinal images and intraocular cytokine profiles remains largely unexplored, and no prior work has systematically compared fundus- and OCT-based deep learning models for cytokine prediction. We aimed to predict intraocular cytokine concentrations using color fundus photographs (CFP) and retinal optical coherence tomography (OCT) with deep learning. Our pipeline consisted of image preprocessing, convolutional neural network–based feature extraction, and regression modeling for each cytokine. Deep learning was implemented using AutoGluon, which automatically explored multiple architectures and converged on ResNet18, reflecting the small dataset size. Four approaches were tested: (1) CFP alone, (2) CFP plus demographic/clinical features, (3) OCT alone, and (4) OCT plus these features. Prediction performance was defined as the mean coefficient of determination (R²) across 34 cytokines, and differences were evaluated using paired two-tailed t-tests. We used data from 139 patients (152 eyes) and 176 aqueous humor samples. The cohort consisted of 85 males (61%) with a mean age of 73 (SD 9.8). Diseases included 64 exudative age-related macular degeneration, 29 brolucizumab-associated endophthalmitis, 19 cataract surgeries, 15 retinal vein occlusion, and 8 diabetic macular edema. Prediction performance was generally poor, with mean R² values below zero across all approaches. The CFP-only model (–0.19) outperformed CFP plus demographics (–24.1; p = 0.0373), and the OCT-only model (–0.18) outperformed OCT plus demographics (–14.7; p = 0.0080). No significant difference was observed between CFP and OCT (p = 0.9281). Notably, VEGF showed low predictability (31st with CFP, 12th with OCT). Full article

Accurate wheat yield prediction is essential for ensuring food security and supporting governmental decision-making. However, the scarcity of long-term agricultural time-series data and the complex interplay between meteorological and socio-economic factors pose significant challenges. To address these issues, this study proposes a Transfer-Learning-Based Parallel CNN–LSTM–Attention (TPCLA) model for wheat yield forecasting. A cross-regional transfer learning strategy is employed to mitigate data scarcity by leveraging temporal patterns learned from regions with similar ecological characteristics. The proposed parallel architecture integrates one-dimensional convolutional neural networks and long short-term memory networks to jointly extract spatial and temporal features, while an attention mechanism is incorporated to highlight key influencing factors and enhance feature interpretability. Unlike conventional studies that primarily focus on climatic variables, this work considers both direct factors (e.g., average temperature and precipitation) and indirect socio-economic factors (e.g., agricultural mechanization level, total agricultural output value, grain production scale, cultivated land area, and disaster-affected area). Experimental results on multivariate wheat data from 1993 to 2024 demonstrate that several indirect indicators exert a more substantial influence on yield than traditional meteorological variables—reflecting the increasing ability of modern agricultural practices to buffer climatic variability. The proposed TPCLA model achieves an RMSE of 0.394, MAE of 0.326, and an $R^2$ of 0.904, outperforming multiple benchmark models and confirming its robustness and predictive superiority under small-sample conditions. The findings not only validate the effectiveness of integrating indirect yield-influencing factors but also provide new insights for agricultural policy formulation and climate resilience strategies. Full article

This paper addresses semi-supervised anomaly detection in settings where only a small subset of normal data can be labeled. Such conditions arise, for example, in industrial quality control of windshield wiper noise, where expert labeling is costly and limited. Our objective is to learn a one-class decision boundary that leverages the geometry of unlabeled data while remaining robust to contamination and scarcity of labeled normals. We propose a graph-attention-regularized deep support vector data description (GAR-DSVDD) model that combines a deep one-class enclosure with a latent k-nearest-neighbor graph whose edges are weighted by similarity- and score-aware attention. The resulting loss integrates (i) a distance-based enclosure on labeled normals, (ii) a graph smoothness term on squared distances over the attention-weighted graph, and (iii) a center-pull regularizer on unlabeled samples to avoid over-smoothing and boundary drift. Experiments on a controlled simulated dataset and an industrial windshield wiper acoustics dataset show that GAR-DSVDD consistently improves the F1 score under scarce label conditions. On average, F1 increases from 0.78 to 0.84 on the simulated benchmark and from 0.63 to 0.86 on the industrial case study relative to the best competing baseline. Full article