Search Results (2,519)

Diplodia shoot blight is an opportunistic fungal pathogen infesting many conifer species and it has a global distribution. Depending on the duration and severity of the disease, affected needles appear yellow (chlorotic) for a brief period before becoming red or brown in colour. These symptoms can occur on individual branches or over the entire crown. Aerial sketch-mapping or the manual interpretation of aerial photography for tree health surveys are labour-intensive and subjective. Recently, however, the application of deep learning (DL) techniques to detect and classify tree crowns in high-spatial-resolution imagery has gained significant attention. This study evaluated two complementary DL approaches for the detection and classification of Pinus radiata trees infected with diplodia shoot blight across five geographically dispersed sites with varying topographies over two acquisition years: (1) object detection using YOLOv12 combined with Segment Anything Model (SAM) and (2) pixel-level semantic segmentation using U-Net, SegFormer, and EVitNet. The three damage classes for the object detection approach were ‘yellow’, ‘red-brown’ (both whole-crown discolouration) and ‘dead tops’ (partially discoloured crowns), while for the semantic segmentation the three classes were yellow, red-brown, and background. The YOLOv12m model achieved an overall mAP50 score of 0.766 and mAP50–95 of 0.447 across all three classes, with red-brown crowns demonstrating the highest detection accuracy (mAP50: 0.918, F1 score: 0.851). For semantic segmentation models, SegFormer showed the strongest performance (IoU of 0.662 for red-brown and 0.542 for yellow) but at the cost of longest training time, while EVitNet offered the most cost-effective solution achieving comparable accuracy to SegFormer but with a superior training efficiency with its lighter architecture. The accurate identification and symptom classification of crown damage symptoms support the calibration and validation of satellite-based monitoring systems and assist in the prioritisation of ground-based diagnosis or management interventions. Full article

Medical image segmentation acts as a foundational element of medical image analysis. Yet its accuracy is frequently limited by the scale fluctuations of anatomical targets and the intricate contextual traits inherent in medical images—including vaguely defined structural boundaries and irregular shape distributions. To tackle these constraints, we design a multi-scale feature attention network (MFA-Net), customized specifically for thyroid nodule, skin lesion, and breast lesion segmentation tasks. This network framework integrates three core components: a Bidirectional Feature Pyramid Network (Bi-FPN), a Slim-neck structure, and the Convolutional Block Attention Module (CBAM). CBAM steers the model to prioritize boundary regions while filtering out irrelevant information, which in turn enhances segmentation precision. Bi-FPN facilitates more robust fusion of multi-scale features via iterative integration of top-down and bottom-up feature maps, supported by lateral and vertical connection pathways. The Slim-neck design is constructed to simplify the network’s architecture while effectively merging multi-scale representations of both target and background areas, thus enhancing the model’s overall performance. Validation across four public datasets covering thyroid ultrasound (TNUI-2021, TN-SCUI 2020), dermoscopy (ISIC 2016), and breast ultrasound (BUSI) shows that our method outperforms state-of-the-art segmentation approaches, achieving Dice similarity coefficients of 0.955, 0.971, 0.976, and 0.846, respectively. Additionally, the model maintains a compact parameter count of just 3.05 million and delivers an extremely fast inference latency of 1.9 milliseconds—metrics that significantly outperform those of current leading segmentation techniques. In summary, the proposed framework demonstrates strong performance in thyroid, skin, and breast lesion segmentation, delivering an optimal trade-off between high accuracy and computational efficiency. Full article

The seeding technique is the only way to precisely control the crystal orientation of single-crystal superalloy castings. However, an inevitable assembly gap exists between the seed and the mold cavity in practice, whose role in defect formation remains insufficiently understood. To elucidate the mechanism and impact of this gap, superalloy seeds were machined to different extents, aiming to create varying gaps with the mold. After the seeding experiment, the chilled layers formed on the perimeter of the pre-processed seeds were detected, exhibiting two distinct microstructural zones: a eutectic aggregation region at the bottom and an equiaxed grain at the top. The thicker the layer, the more pronounced the differences in microstructure between these two regions. This can be explained by the fact that during preheating, the γ/γ′ eutectic-rich interdendritic region (enriched with Al + Ti + Ta) in the original seed melted first due to its lower melting point. The molten fluid flowed downward into the gap, solidifying rapidly into the chilled layer. The leading portion of the fluid, melting from the interdendritic zone, formed the eutectic zone in the lower part of the chilled layer. The subsequently poured charge alloy melt (non-enriched with Al + Ti + Ta) generated the upper equiaxed zone with only a little γ/γ′ eutectic. These equiaxed grains in the chilled layer subsequently grew upward and potentially developed into stray grains of the casting. Full article

The analysis of ground-penetrating radar data generally relies on the visual identification of structures on selected profiles and their interpretation in terms of buried features. In simple cases, inverse modelling of the acquired data set can facilitate interpretation and reduce subjectivity. These methods suffer from severe restrictions due to antenna resolution limits, which prevent the identification of tiny structures, particularly in forensic, stratigraphic, and engineering applications. Here, we describe a technique to obtain a high-resolution characterization of the underground, based on the forward modelling of individual traces (A-scans) of selected radar profiles. The model traces are built by superposition of Ricker wavelets with different polarities, amplitudes, and arrival times and are used to create reflectivity diagrams that plot reflection amplitudes and polarities versus depth. A thin bed is defined as a layer of higher or lower permittivity relative to the surrounding material, such that the top and bottom reflections are subject to constructive interference, determining the formation of an anomalous peak in the trace (tuning effect). The proposed method allows the detection of ultra-thin layers, well beyond the Rayleigh vertical resolution of GPR antennas. This approach requires a preliminary estimation of the instrumental uncertainty of common monostatic antennas and takes into account the frequency-dependent attenuation, which causes a spectral shift of the dominant frequency acquired by the receiver antenna. Such a quantitative approach to analyzing radar data can be used in several applications, notably in stratigraphic, forensic, paleontological, civil engineering, heritage protection, and soil stratigraphy applications. Full article

Semi-supervised learning has attracted widespread attention due to its ability to utilize both labeled and unlabeled data, leading to significant progress in recent years. Conventional semi-supervised learning approaches often rely on a strategy that combines weak and strong image-level augmentations and employs pseudo-labeling techniques, where high-confidence predictions are selected as pseudo-labels while low-confidence ones are discarded. However, such methods tend to overlook the useful information contained in low-confidence samples. Moreover, existing augmentation strategies are mostly limited to the image level and lack feature-level perturbations. To address these limitations, this paper proposes a semi-supervised learning method that integrates complementary labels and auxiliary feature perturbations, aiming to extract valuable information from low-confidence samples and expand the scope of feature-space perturbations. Experiments on the standard CIFAR-10 dataset show that under the extreme setting of only 4 labels per class, our method improves accuracy by 2.58% compared to Fixmatch. On the more complex STL-10 dataset, also with only 4 labels per class, the Top-1 accuracy is improved by 2.59%. In addition, systematic ablation studies are conducted to verify the effectiveness of each component. Full article

Prediction of stock closing price plays a critical role in financial planning, risk management, and informed investment decision-making. In this study, we propose a novel model that synergistically amalgamates Bidirectional GRU (BiGRU) with three complementary attention techniques—Top-k Sparse, Global, and Bahdanau Attention—to tackle the complex, intricate, and non-linear temporal dependencies in financial time series. The proposed Fused Attention Model is validated on two highly volatile, non-linear, and complex- patterned stock indices: NIFTY 50 and S&P 500, with 80% of the historical price data used for model learning and the remaining 20% for testing. A comprehensive analysis of the results, benchmarked against various baseline and hybrid deep learning architectures across multiple regression performance metrics such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and R² Score, demonstrates the superiority and noteworthiness of our proposed Fused Attention Model. Most significantly, the proposed model yields the highest prediction accuracy and generalization capability, with R² scores of 0.9955 on NIFTY 50 and 0.9961 on S&P 500. Additionally, to mitigate the issues of interpretability and transparency of the deep learning model for financial forecasting, we utilized three different Explainable Artificial Intelligence (XAI) techniques, namely Integrated Gradients, SHapley Additive exPlanations (SHAP), and Attention Weight Analysis. The results of these three XAI techniques validated the utilization of three attention techniques along with the BiGRU model. The explainability of the proposed model named as BiGRU based Fused Attention (BiG-FA), in addition to its superior performance, thus offers a robust and interpretable deep learning model for time-series prediction, making it applicable beyond the financial domain. Full article

Large language models (LLMs) have rapidly transformed artificial intelligence applications across industries, yet their integration into production systems has unveiled critical security vulnerabilities, chief among them prompt injection attacks. This comprehensive review synthesizes research from 2023 to 2025, analyzing 45 key sources, industry security reports, and documented real-world exploits. We examine the taxonomy of prompt injection techniques, including direct jailbreaking and indirect injection through external content. The rise of AI agent systems and the Model Context Protocol (MCP) has dramatically expanded attack surfaces, introducing vulnerabilities such as tool poisoning and credential theft. We document critical incidents including GitHub Copilot’s CVE-2025-53773 remote code execution vulnerability (CVSS 9.6) and ChatGPT’s Windows license key exposure. Research demonstrates that just five carefully crafted documents can manipulate AI responses 90% of the time through Retrieval-Augmented Generation (RAG) poisoning. We propose PALADIN, a defense-in-depth framework implementing five protective layers. This review provides actionable mitigation strategies based on OWASP Top 10 for LLM Applications 2025, identifies fundamental limitations including the stochastic nature problem and alignment paradox, and proposes research directions for architecturally secure AI systems. Our analysis reveals that prompt injection represents a fundamental architectural vulnerability requiring defense-in-depth approaches rather than singular solutions. Full article

Structural health monitoring (SHM) increasingly contributes to the safety and durability of key infrastructure, especially bridges. This research introduces a rotation-based approach for damage detection and quantification using a damage candidate grid search technique (DCGS) on simply supported girder bridges under quasi-static or slowly moving loading conditions. Applying the principle of virtual work, the healthy and candidate-damaged rotation responses are analytically obtained and compared with the rotation observed directly at the moving load location. Damage is defined in terms of three key parameters: the start and the end of the damage, $L_1$ and $L_2$, respectively, and the damage severity $\beta$. The DCGS method is validated using finite element model simulations of 12 damage scenarios subjected to different noise levels. A statistical analysis and confidence interval characterize the accuracy and consistency of the top ten estimations produced by the DCGS method. A damage length ratio (DLR), defined from the span of the beam, $L$, and the damage location, $L_1$ and $L_2$, improves the robustness of the methodology against measurement noise by reducing possible false positive estimations. Additionally, the experimental results on two beam structures further validate the method. Absolute relative errors (AREs) of about 6% and absolute errors (AEs) of around 0.16 between the estimated and real damage parameters characterize the performance of the technique, considering damage location and damage severity, respectively. The results show that the DCGS methodology can effectively locate damage and estimate its severity in the presence of noise. The developed framework provides a sensitive and practical SHM tool that is suitable for early damage detection in railway and road bridges. Full article

Low clouds significantly impact weather, climate, and multiple environmental and economic sectors such as agriculture, fire risk management, aviation, and renewable energy. Accurate knowledge of cloud base height (CBH) is critical for optimizing crop yields, improving fire danger forecasts, enhancing flight safety, and increasing solar energy efficiency. This study evaluates a shadow-based CBH retrieval method using Moderate Resolution Imaging Spectroradiometer (MODIS) satellite visible imagery and compares the results against collocated lidar measurements from the Micro-Pulse Lidar Network (MPLNET) ground stations. The shadow method leverages sun–sensor geometry to estimate CBH from the displacement of cloud shadows on the surface, offering a practical and high-resolution passive remote sensing technique, especially useful where active sensors are unavailable. The validation results show strong agreement, with a correlation coefficient (R) = 0.96 between shadow-based and lidar-derived CBH estimates, confirming the robustness of the approach for shallow, isolated cumulus clouds. The method’s advantages include direct physical height estimation without reliance on cloud top heights or stereo imaging, applicability across archived datasets, and suitability for diurnal studies. This work highlights the potential of shadow-based retrievals as a reliable, cost-effective tool for global low cloud monitoring, with important implications for atmospheric research and operational forecasting. Full article

In inland waters, Atmospheric Correction (AC), including Adjacency Effect (AE) correction, is a major challenge for water quality retrieval using optical satellite data. This study evaluated three image pre-processing options for turbidity retrieval in the St. Lawrence River using Sentinel-2 (S2) and Landsat 8/9 (L8/9) imagery with the Light Gradient Boosting Machine (LightGBM) model: (1) No pre-processing, i.e., use of Top-of-Atmosphere (TOA) reflectance, (2) AC pre-processing, obtaining water-leaving reflectance (R_w) from AC for the Operational Land Imager lite (ACOLITE)’s Dark Spectrum Fitting (DSF) technique, and (3) AE pre-processing, correcting for the AE using T-Mart before obtaining R_w from DSF. Results demonstrated that AE pre-processing outperformed the other two options. For L8/9, AE pre-processing reduced the Root Mean Square Error (RMSE) and improved the median symmetric accuracy (ε) by 48.8% and 19.0%, respectively, compared with AC pre-processing, and by 48.5% and 50.7%, respectively, compared with No pre-processing. For S2, AE pre-processing performed better than AC pre-processing and also outperformed No pre-processing, reducing RMSE by 28.4% and ε by 50.8%. However, No pre-processing yielded the lowest absolute symmetric signed percentage bias (|β|) among all pre-processing options. Analysis indicated that AE pre-processing yielded superior performance within 0–300 m from shore than other options, where the AE influence is strongest. Turbidity maps generated using AE pre-processing were smoother and less noisy compared to the other pre-processing options, particularly in cloud-adjacent regions. Overall, our findings suggest that incorporating AE correction through T-Mart improves the performance of the LightGBM model for turbidity retrieval from both L8/9 and S2 imagery in the St. Lawrence River, compared to the alternative pre-processing options. Full article

Presbyopia is a physiological phenomenon and one of the leading factors contributing to decreased near visual acuity. The prevalence of presbyopia, its social and economic consequences and the prolongation of human life place the correction of presbyopia among the top challenges in modern ophthalmology. Despite the numerous methods currently available for correcting presbyopia, there is still no ideal technique that, by restoring the eye’s age-related loss of physiological accommodation, would provide long-term effectiveness without adverse effects. This article offers an overview of the existing knowledge on the etiology of presbyopia and the available methods of its correction, with particular emphasis on refractive surgery techniques. Full article

Assurance of the integrity of every weld joint is highly desirable, and defect detection methods that can be applied to welds at high temperatures immediately after welding are required. The laser ultrasonic (LU) method, which generates ultrasonic waves in the target via pulsed laser irradiation, is a well-known technique for non-contact defect detection during welding. Ultrasonic waves excited in ablation mode exhibit large amplitudes and predominantly surface-normal propagation, which has driven extensive research into their application for weld inspection. However, owing to the size and weight of conventional equipment, such systems have largely been limited to bench-top experimental setups. To address this, we developed an LU robotic system incorporating a compact, lightweight laser source and an improved signal-processing system. We conducted experiments to measure signals and to detect backside slits in flat plates and blowholes in lap-fillet welds. Additionally, a method to improve the sensitivity of laser interferometers was investigated and demonstrated on smut-covered areas near weld beads. Full article

The miniaturisation of electronic devices has intensified the demand for compact, high-performance lithium-ion batteries. This review synthesises recent progress in microscale battery development, focusing on microfabrication techniques, nanostructured materials, porosity-engineered architectures, and strategies for reducing non-active components. It explores both top–down and bottom–up fabrication methods, the integration of nanomaterials, the role of gradient electrode architectures in enhancing ion transport and energy density, along with strategies to reduce non-active components, such as separators and current collectors, to maximise volumetric efficiency. Advances in top–down and bottom–up fabrication methods, including photolithography, laser structuring, screen printing, spray coating, mechanical structuring, and 3D printing, enable precise control over electrode geometry and enhance ion transport and material utilisation. Nanostructured anodes, cathodes, electrolytes, and separators further improve conductivity, mechanical stability, and cycling performance. Gradient porosity designs optimise ion distribution in thick electrodes, while innovations in ultra-thin separators and lightweight current collectors support higher energy density. Remaining challenges relate to scalability, mechanical robustness, and long-term stability, especially in fully integrated micro-battery architectures. Future development will rely on hybrid fabrication methods, advanced material compatibility, and data-driven optimisation to bridge laboratory innovations with practical applications. By integrating microfabrication and nanoscale engineering, next-generation LIBs can deliver high energy density and long operational lifetimes for miniaturised and flexible electronic systems. Full article

Speech is one of the earliest-affected modalities in Parkinson’s disease (PD). For more reliable PD evaluation, speech-based telediagnosis studies often use multiple samples from the same subject to capture variability in speech recordings. However, many existing studies split samples—rather than subjects—between training and testing, creating a biased experimental setup that allows data (samples) from the same subject to appear in both sets. This raises concerns for reliable PD evaluation due to data leakage, which results in over-optimistic performance (often close to 100%). In addition, detecting subtle vocal impairments from speech recordings using multiple feature extraction techniques often increases data dimensionality, although only some features are discriminative while others are redundant or non-informative. To address this and build a reliable speech-based PD telediagnosis system, the key contributions of this work are two-fold: (1) a uniform (fair) experimental setup employing subject-wise symmetric (stratified) splitting in 5-fold cross-validation to ensure better generalization in PD prediction, and (2) a novel hybrid data complexity-aware (HDC) feature selection method that improves class separability. This work further contributes to the research community by releasing a publicly accessible five-fold benchmark version of the Parkinson’s speech dataset for consistent and reproducible evaluation. The proposed HDC method analyzes multiple aspects of class separability to select discriminative features, resulting in reduced data complexity in the feature space. In particular, it uses data complexity measures (F4, F1, F3) to assess minimal feature overlap and ReliefF to evaluate the separation of borderline points. Experimental results show that the top-50 discriminative features selected by the proposed HDC outperform existing feature selection algorithms on five out of six classifiers, achieving the highest performance with 0.86 accuracy, 0.70 G-mean, 0.91 F1-score, and 0.58 MCC using an SVM (RBF) classifier. Full article

Modern smart cities face increasing pressure to invest in sustainable and reliable energy systems while navigating uncertainties arising from fluctuating market conditions, evolving technology landscapes, and diverse expert opinions. Traditional multi-criteria decision-making (MCDM) approaches often fail to fully represent these uncertainties as they typically rely on crisp inputs, lack temporal memory, and do not explicitly account for stochastic variability. To address these limitations, this study introduces a novel Stochastic Fractional Fuzzy Tensor (SFFT)-based Group Decision-Making framework. The proposed approach integrates three dimensions of uncertainty within a unified mathematical structure: fuzzy representation of subjective expert assessments, fractional temporal operators (Caputo derivative, $\alpha =0.85$) to model the influence of historical evaluations, and stochastic diffusion terms ($\sigma =0.05$) to capture real-world volatility. A complete decision algorithm is developed and applied to a realistic smart city renewable energy selection problem involving six alternatives and six criteria evaluated by three experts. The SFFT-based evaluation identified Geothermal Energy as the optimal choice with a score of 0.798, followed by Offshore Wind (0.722) and Waste-to-Hydrogen (0.713). Comparative evaluation against benchmark MCDM methods—TOPSIS (Technique for Order Preference by Similarity to Ideal Solution), VIKOR (VIšekriterijumsko KOmpromisno Rangiranje), and WSM (Weighted Sum Model)—demonstrates that the SFFT approach yields more robust and stable rankings, particularly under uncertainty and model perturbations. Extensive sensitivity analysis confirms high resilience of the top-ranked alternative, with Geothermal retaining the first position in 82.4% of 5000 Monte Carlo simulations under simultaneous variations in weights, memory parameter ($\alpha \in [0.25,0.95]$), and noise intensity ($\sigma \in [0.01,0.10]$). This research provides a realistic, mathematically grounded, and decision-maker-friendly tool for strategic planning in uncertain, dynamic urban environments, with strong potential for deployment in wider engineering, management, and policy applications. Full article