Search Results (781)

LiDAR point clouds are widely used in remote sensing perception scenarios, such as autonomous driving. However, LiDAR-based perception models remain vulnerable to adversarial perturbations, which may compromise the reliability of safety-critical 3D perception systems. Among different attack paradigms, transfer-based attacks are particularly practical because they generate adversarial examples on accessible surrogate models and apply the generated examples directly to unknown target models. Nevertheless, existing transferable attacks on point clouds often perturb regions that are discriminative for the surrogate model but insufficiently stable across different architectures, leading to limited transferability and noticeable geometric distortion. To address this problem, we propose SAGE, a Saliency And Geometry Enhanced transferable attack framework for LiDAR point cloud perception in remote sensing. Specifically, SAGE unifies point-coordinate priors with source-model gradient signals to generate a saliency map, which serves as a transferable indicator of vulnerable local structures. SAGE further leverages this map through saliency-guided perturbation allocation and explicit geometric constraints to enhance transferability while preserving point-cloud geometry. To demonstrate the effectiveness of SAGE, we evaluate SAGE on point-cloud classification benchmarks and further validate it on LiDAR-based 3D object detection using KITTI and nuScenes. Experimental results show that SAGE consistently outperforms existing transferable attack methods in attack success rate while preserving favorable geometric quality of adversarial point clouds. These findings demonstrate that SAGE offers an effective and practical framework for assessing the transfer robustness of LiDAR-based remote sensing perception systems. Full article

The performance of haul trucks in open-pit mining is strongly affected by haul-road geometry, surface condition, rolling resistance, and operational traffic regimes. However, existing studies often consider road-surface mapping, vehicle dynamic response, and onboard telemetry as separate information streams, which limits the reproducible assessment of how road-related factors are associated with VIMS-derived fuel-use proxy burden, mechanical dynamic response, and transport-cycle instability. This study proposes a field-based, segment-level multisensor framework that integrates unmanned aerial vehicle/light detection and ranging (UAV/LiDAR) road-surface reconstruction, global positioning system/inertial measurement unit (GPS/IMU) trajectory and vibration data, and Caterpillar Vial Information Management System (VIMS) telemetry into a unified spatiotemporal analytical dataset. The methodological contribution consists in the synchronization of heterogeneous data sources at the road-segment level, the calculation of interpretable road-condition and vehicle-response indicators, and the statistical assessment of road-related effects while explicitly accounting for confounding factors such as longitudinal grade, payload state, speed regime, truck class, and operational variability. Unlike studies that use LiDAR mapping, vibration monitoring, or onboard telemetry as separate diagnostic channels, the proposed approach introduces a segment-level analytical framework in which road morphology, truck response, and operational penalties are aligned within the same spatial unit, interpreted under confounder-aware conditions, and verified through repeat-pass reproducibility and robustness checks. The framework was tested on haul roads around the Ekibastuz open-pit coal mine. The field analysis identifies road segments where degraded surface morphology, increased waviness, unfavorable longitudinal profile, and higher rolling resistance coincide with increased mechanical dynamic response, VIMS-derived fuel-use proxy burden, braking instability, and travel-time variability. The results are interpreted as controlled field-supported associations rather than as isolated causal effects. The proposed maintenance ranking should therefore be regarded as a decision-support output, while the operational effectiveness of specific repair interventions requires future before–after validation. Full article

This study proposes a deep learning-based framework for the automated classification of photovoltaic solar cells as defective or normal using electroluminescence (EL) imaging. A balanced dataset containing 20,400 EL images, comprising 10,200 defective and 10,200 normal solar cells, was used for model development and evaluation. To reflect practical inspection requirements, cracked and broken cells were combined into a single defective category, resulting in a binary classification task. The dataset includes both monocrystalline and polycrystalline solar cells, which were analyzed together within a unified classification framework to improve applicability to real-world photovoltaic systems. To ensure a fair and unbiased evaluation, dataset partitioning was performed prior to any preprocessing or augmentation operations, and each image was assigned exclusively to the training, validation, or test subset. Data augmentation was applied only to the training set, eliminating the possibility of data leakage. Four state-of-the-art deep learning architectures, EfficientNet-B2, ConvNeXt-Tiny, MaxViT-T, and ResNet-50, were trained and evaluated under identical experimental conditions using the same preprocessing pipeline, training strategy, and dataset split. Model performance was assessed using accuracy, precision, recall, F1-score, ROC-AUC, confusion matrices, and explainability-based activation and attention heat maps. All evaluated models achieved classification accuracies exceeding 98%, demonstrating strong capability for EL-based defect detection. EfficientNet-B2 achieved the highest numerical performance, reaching 99.31% accuracy, 0.9931 F1-score, and 0.9987 ROC-AUC. MaxViT-T exhibited similarly strong performance with rapid convergence and balanced class-wise metrics, while ConvNeXt-Tiny and ResNet-50 also produced highly reliable results. Heat map visualizations revealed that EfficientNet-B2 and MaxViT-T concentrated their attention more precisely on defect regions such as cracks and fractures, providing visual interpretability in addition to quantitative performance. The results demonstrate that modern deep learning architectures can accurately and reliably detect photovoltaic cell defects from EL images under a unified binary classification framework. Furthermore, explainability techniques enhance the transparency of model predictions, supporting the practical deployment of intelligent inspection systems for photovoltaic manufacturing and maintenance applications. Full article

As a pilot-scale feasibility study, aging bridge infrastructure and limited inspection resources have created an urgent need for automated and reliable bridge condition assessment systems. Most existing deep learning-based inspection approaches detect damage types from images without considering the structural member on which the damage occurs, limiting their practical utility for maintenance decision-making. This study proposes a structure-aware deep learning framework for automated bridge inspection that integrates structural member segmentation, two-class damage detection, and spatial damage-to-member association within a unified pipeline. A SegFormer-based semantic segmentation model was trained on a custom bridge inspection dataset comprising 1339 images to identify three primary structural member classes—main girder, deck slab, and abutment—achieving a test mean Intersection over Union (mIoU) of 0.851. Boundary refinement using the Segment Anything Model (SAM) in mask-prompt mode was applied to improve mask precision during training data preparation. A YOLOv8s object detection model was trained on a custom bridge damage dataset of 9142 annotated images (6531 training, 1740 validation, and 871 test images) to detect two damage classes—crack and corrosion—achieving a mean Average Precision (mAP50) of 0.445 at a confidence threshold of 0.30. The framework associates detected damage with segmented structural members using a region-based spatial assignment strategy, enabling structure-aware outputs such as “crack on main girder” and “corrosion on deck slab.” Manual evaluation on 100 bridge inspection images demonstrated a fully correct damage detection accuracy of 70.0% and a fully correct member assignment accuracy of 62.0%. When partially correct predictions were additionally considered for qualitative analysis, the corresponding accuracies increased to 84.0% and 87.0%, respectively. The main girder class achieved the highest combined accuracy for both damage detection (90.9%) and member assignment (93.9%). These results demonstrate the potential of the proposed framework as a first layer for AI-assisted bridge inspection by associating detected damage with structural members, providing structured inspection information to support subsequent maintenance assessment and infrastructure monitoring. Full article

Reliable and scalable landmine detection technologies are essential for humanitarian mine action (HMA), yet standardized benchmarks for Unmanned Aerial Vehicle (UAV)-based sensing in operationally relevant environments remain limited. This study presents a comprehensive evaluation of 34 multimodal datasets acquired over a standardized seeded test site for landmine and unexploded ordnance detection. Nine sensing modalities, including RGB, thermal, multispectral, hyperspectral, LiDAR, and Synthetic Aperture Radar (SAR), are evaluated using the Anomaly, Identifiable Anomaly, Unique Identifiable Anomaly (AIU) index to establish a unified framework for quantifying detection fidelity. Results indicate that RGB imagery achieves the highest surface detection rate (94.8%), with 45.4% of targets classified as uniquely identifiable, reducing false-positive risk. For sub-surface detection, handheld electromagnetic induction (EMI) and magnetometry exceed 95% detection for ferrous items but fall below 10% for plastic ordnance. Ground-penetrating radar (GPR) is the only modality capable of detecting buried plastic targets (55.6% for cart-based systems), whereas UAV-mounted GPR remains limited (18.2%) at current operational flight heights. Based on the comparative analysis, we discuss the gaps in current detection capabilities, compare false-positive rates across modalities, and perform a cost–benefit analysis fitting contamination scenarios with the most cost-effective detection method. All datasets are publicly released, along with an interactive web-map, to support reproducible benchmarking and cross-modality comparison in UAV-enabled explosive hazard detection. Full article

Accurate object detection in agricultural environments remains challenging due to illumination variation, background clutter, partial occlusion, and overlapping fruits. Conventional object detection methods mainly rely on deterministic data augmentation strategies or feature-level refinement, which often exhibit limited robustness under complex field conditions. To address this issue, this paper proposes a Diffusion-Regularized Object Detection (DROD) framework for robust pineapple target detection in agricultural imagery. The proposed framework introduces a mathematically grounded forward diffusion and diffusion-guided representation mechanism directly in the image domain, where stochastic perturbations are generated through forward diffusion and semantically meaningful image representations are learned via diffusion-guided representation. A unified optimization framework and theoretical analyses of perturbation propagation, Lipschitz stability, and training convergence are further established to provide mathematical support for the proposed method. Extensive experiments were conducted on a self-constructed dataset containing 1600 real-world pineapple images collected under practical agricultural conditions. Comparative evaluations involving YOLOv8-s, YOLOv8-L, traditional data augmentation, and the recent JTA:GAN method demonstrate that the proposed DROD framework consistently achieves the best detection performance in terms of Precision, Recall, mAP@0.5, and mAP@0.5:0.95 while maintaining computational complexity and inference speed comparable to the original YOLOv8 architecture. Furthermore, ablation studies, diffusion parameter sensitivity analysis, visualization analysis, and experimental validation under different perturbation levels consistently verify the effectiveness and robustness of the proposed diffusion mechanism. These results demonstrate that diffusion-based regularization provides an effective and computationally efficient solution for robust agricultural object detection and offers a practical framework for intelligent precision agriculture applications. Full article

Artificial intelligence is increasingly applied to earthquake-related seismo-ionospheric analysis with total electron content (TEC), but whether this literature is converging methodologically remains unresolved. We conducted a mapping review of 56 English-language journal articles retrieved from Scopus and Web of Science to characterize how artificial intelligence and computational intelligence methods are used with TEC in seismo-ionospheric and multi-precursor frameworks. The corpus shows recent growth in scientific production, strong concentration in a limited set of countries, institutions, and journals, and a stable conceptual backbone centered on earthquake, ionosphere, TEC, GPS-TEC, precursors, prediction-related terminology, anomaly detection, machine learning, and deep learning. However, full-text synthesis of the included studies shows that this thematic coherence coexists with substantial methodological divergence. We identified a transition from classical TEC anomaly detection toward AI-assisted decision systems, including models that forecast expected TEC behavior, flag candidate anomalies, classify precursor-like or disturbance-related states, and support monitoring-oriented outputs. We also identified a distinct operational strand focused on near-real-time detection of coseismic and tsunami-related ionospheric disturbances rather than deterministic earthquake prediction. Across these formulations, anomaly definitions, TEC representations, confounder control, baselines, uncertainty handling, and validation strategies remain pipeline-dependent, which limits cumulative comparability and physical interpretability across studies. These findings indicate that the field is thematically focused but not yet methodologically unified. Future progress will depend less on adding isolated case studies and more on clearer anomaly criteria, stronger control of solar and geomagnetic effects, explicit baselines, event-wise and region-wise validation, systematic false-alarm reporting, uncertainty-aware outputs, and transparent documentation of preprocessing and modeling decisions. Full article

Accurate and transferable estimation of forest carbon is essential for operational forest management and national greenhouse gas reporting, yet it remains challenging because of variation in stand structure and site conditions. Unmanned aerial vehicle-based light detection and ranging (UAV-LiDAR) provides detailed structural information for modelling tree-level carbon, but model transferability across sites is often limited. In this study, we compared three modelling approaches—a linear mixed-effects model (LMM), a generalised additive model (GAM), and Random Forest (RF)—within a unified framework of multi-site, locally post hoc calibrated, and fully local model-fitting strategies. Using data from 20 radiata pine (Pinus radiata D. Don) plantation stands across New Zealand (35,201 trees), a leave-one-site-out (LOSO) framework was used to assess multi-site model transferability and support post hoc calibration, while local models were evaluated using repeated within-site train/test splits. We also evaluated how prediction accuracy changed with increasing local sample size and compared random tree selection with plot-based sampling. Multi-site models showed poor generalisation, with mean relative RMSE ranging from 35.9% to 56.9% and substantial site-level bias. Applying post hoc calibration to the multi-site model using a 50-tree sample reduced prediction error by 30 to 60% (mean relative RMSE 22.8–25.0%) and substantially reduced bias across sites. The fitting of fully local models with the same sample size yielded only modest further improvements (mean relative RMSE 21.9–23.1%). Gains in accuracy were minimal with increasing sample sizes above 50 trees for post hoc calibration and 175 trees for the fully local models, and differences in accuracy between sampling strategies were small. These results show that post hoc calibration of multi-site UAV-LiDAR models with a small local sample provides a practical and efficient approach for tree-level carbon estimation in plantation forests. Full article

Insider threats continue to be the most persistent and most destructive threat to cybersecurity; malicious or negligent users work only in the real-time restricted area of the organization and are gradually breaking the boundaries of company norms. Conventional rule-based and statistical detection methods have difficulty detecting inconspicuous, context-dependent, and ever-changing behavior, leading to detection delays and high false-positive rates. Our paper introduces an explainable AI-based Insider-Threat Detection (AIB-ITD) model that integrates enterprise telemetry—including email, web, logon/VPN, and file events—into a unified behavioral framework. The effectiveness of combining heterogeneous behavioral indicators observed in AIB-ITD is consistent with recent behavioral analytics implementations that have demonstrated the value of multimodal user-behavior profiling for insider-threat identification in enterprise environments. The proposed AIB-ITD framework is based on anomaly-driven processing, unsupervised models (Isolation Forest, PCA reconstruction, and Autoencoder) are combined with sequential modeling (with an LSTM Autoencoder) to model both static and temporal deviations in behavior. An ensemble strategy is applied to combine the outputs of these models to yield a probabilistic insider risk score. To improve transparent analysis and to help the analyst gain trust, SHapley Additive Explanations (SHAP) is used to keep every detection outcome transparent and interpretable using the features. It also integrates feature correlation analysis, static vs sequential-model comparisons, and SHAP stability assessment to validate methodological robustness and reproducibility. An experimental review of the hybrid ensemble using the SEI/CMU CERT Insider Threat Dataset reveals that it performs better than single models for anomaly detection and stability, especially with the inclusion of temporal patterns. The assessment prioritizes anomaly score consistency and reliable risk ranking, rather than classification accuracy, to better reflect real deployment scenarios. In addition, an Automated Response and Privilege Containment (ARPC) feature automatically converts risk scores to multilevel mitigation actions that serve to protect the privacy of the user as the least privileged policies are enforced promptly. The proposed model showed superior robustness, stability, and operational effectiveness to classical methods, especially in the presence of scarce labeled data. Through hybrid anomaly recognition, explainable AI and automated response, AIB-ITD is a practical and scalable solution for next-generation insider-threat detection in enterprise systems. Full article

Mobile inspection robots require reliable license plate detection and recognition under constrained computing resources, small-scale or distant imaging conditions, motion blur, and complex background interference. To address these coupled challenges, this paper proposes a lightweight robot-view visual sensing framework for CPU-oriented license plate perception. Instead of simply stacking network modules, the proposed framework follows a unified design principle of reducing redundant computation while compensating for task-critical visual information. In the detection stage, a YOLOv8-MGL detector is developed based on YOLOv8n by combining GhostC2f-based lightweight feature aggregation with LSKAlite-based contextual enhancement after the SPPF module. In the recognition stage, SimAM is embedded into LPRNet to enhance discriminative character responses under motion blur, low resolution, and local degradation without introducing additional learnable parameters. Experiments on the held-out EDRV-LP test set show that YOLOv8-MGL achieves 99.5% mAP₅₀ and 71.1% mAP_50:95, while reducing the number of parameters from 3.01 M to 2.77 M and GFLOPs from 8.1 to 7.5 compared with YOLOv8n. On a CPU-only Intel Xeon Platinum 8260C platform, YOLOv8-MGL achieves 23.98 FPS. SimAM-LPRNet improves the module-level cropped-plate recognition accuracy from 83.10% to 87.17%. To further examine system-level feasibility, a supplementary YOLOv8-MGL + CRNN-CTC pipeline is evaluated from raw images to final plate strings, achieving 91.0% exact recognition accuracy on the held-out EDRV-LP test set, 92.0% on a non-overlapping external CCPD subset, and 13.25 FPS for complete CPU-only processing. These results demonstrate that the proposed framework provides a favorable trade-off among model compactness, localization quality, recognition robustness, and CPU-oriented inference feasibility for mobile robotic inspection scenarios. Full article

Deception detection is a multifaceted challenge that has gained attention in domains such as forensics, security, and human–computer interaction. However, most EEG-based studies focus on binary classification between truthful and deceptive responses, overlooking the complexity of cognitive processes underlying different deceptive strategies. To address this limitation, we present a multi-class EEG dataset designed to investigate distinct behavioral roles in deception, including honest, bluffer, liar, and deceiver, collected from 51 participants using a controlled mock-crime scenario. In this setup, subjects were assigned predefined roles and interrogated under a standardized protocol with carefully designed questions and responses. EEG signals were recorded using a 16-channel Biosemi ActiveTwo system at a sampling rate of 2048 Hz, with event markers enabling precise temporal segmentation of experimental phases. The dataset captures neural activity associated with varying cognitive load and decision-making across deception types. To the best of our knowledge, this is the first EEG dataset that explicitly incorporates and differentiates four distinct deception-related behavioral roles within a unified experimental framework. Full article

Industrial visual inspection relies heavily on unsupervised anomaly detection due to the scarcity of annotated defect samples. However, existing methods struggle to balance global structural consistency and local defect sensitivity, leading to limited accuracy in practical scenarios. To address this challenge, we propose a unified generative–discriminative framework that combines regularized latent space encoding with multi-scale discriminator-guided supervision. Specifically, a dynamic compactness regularization strategy constrains latent representations of normal samples into a compact manifold to suppress anomaly reconstruction, while a multi-scale discriminator provides hierarchical perceptual constraints to enhance fine-grained anomaly localization across different spatial resolutions. Here we show that the proposed method achieves 98.6% image-level and 98.4% pixel-level AUROC on the MVTec AD benchmark, outperforming state-of-the-art approaches. This framework provides a stable and effective solution for real-world industrial quality inspection. Full article

Traditional SLAM methods face significant challenges in complex environments, including high computational complexity, ambiguous data association, and limited real-time performance. Existing approaches often rely on explicit data association or computationally intensive filtering frameworks, which restrict their scalability and robustness. In this paper, we propose a pose-graph-optimization-based Poisson multi-Bernoulli (PMB) SLAM framework. The proposed method models the map as a unified structure consisting of undetected features represented by a Poisson point process (PPP) and detected features represented by multi-Bernoulli (MB) components, enabling consistent feature estimation while reducing the reliance on explicit data association. Furthermore, an information-controlled pose graph constraint selection strategy (IC-PGCS) is developed to effectively couple PMB filtering with pose graph optimization, allowing adaptive activation of graph optimization based on accumulated information. Simulation results demonstrate that the proposed method achieves comparable map feature estimation accuracy while improving computational efficiency and real-time performance compared with RB-PHD-SLAM and multi-Bernoulli-based SLAM methods. These results validate the effectiveness of the proposed framework for SLAM applications in cluttered indoor environments. Full article

This paper proposes a novel data-driven K-Gap metric method based on the Koopman operator for the detection and isolation of multiplicative faults in high-speed train suspension systems. A systematic comparison is conducted with a data-driven K-Gap approach implemented through the fuzzy modeling framework. First, Takagi–Sugeno (T–S) theory is employed to extend the K-Gap metric for nonlinear dynamic modeling of the suspension system. Subsequently, the Koopman operator framework is introduced to lift the system states into a high-dimensional observable space, enabling a globally linear representation of the system. Building upon Koopman-based stable kernel representation (SKR), a more accurate K-Gap residual metric is constructed. Finally, a unified fault diagnosis scheme is developed with the K-Gap metric as the core indicator, and the two approaches are experimentally compared in terms of their performance in detecting and isolating multiplicative faults. The experimental results demonstrate that the Koopman-based method significantly outperforms the T–S fuzzy model in terms of residual separability, fault classification accuracy, and diagnostic stability, confirming its effectiveness and superiority for fault diagnosis in complex nonlinear systems. Full article

Rising atmospheric CO₂ levels pose a critical challenge to achieving global sustainability targets. Geological carbon sequestration (GCS) offers a long-term solution for reducing greenhouse gas emissions, but its large-scale deployment faces limitations in cost, uncertainty, and operational risk. Recent advances in machine learning (ML) present transformative opportunities to enhance every stage of the carbon capture and storage (CCS) lifecycle, from pre-injection evaluation to post-injection monitoring. This review systematically examines ML integration in CCS applications, emphasizing roles in geological characterization, injection optimization, plume prediction, and leakage detection. It provides a structured overview of ML methodologies including Random Forest, Support Vector Regression, and XGBoost, along with emerging deep learning models used for anomaly detection and uncertainty quantification. Experimental insights, monitoring techniques, and real-time data applications are summarized to illustrate ML’s capability in accelerating simulations, reducing costs, and increasing safety assurance. Furthermore, real-world case studies such as Sleipner (Norway), Illinois Basin–Decatur (USA), Boundary Dam (Canada), Gorgon (Australia), and Quest (Canada) demonstrate how ML has enhanced performance, predictive accuracy, and storage reliability in field-scale CCS projects. The review concludes by identifying existing challenges, data scarcity, interpretability, and regulatory integration, and proposes a unified ML framework for scalable, autonomous, and secure CO₂ storage. Overall, this study provides a comprehensive roadmap for leveraging artificial intelligence to achieve reliable, cost-effective, and sustainable carbon management solutions aligned with global net-zero objectives. Full article