Search Results (6,070)

Although Convolutional Neural Networks (CNNs) have delivered state-of-the-art accuracy in plant disease classification, their deployment is still hindered by data scarcity, computational cost, and architectural heterogeneity. Transfer learning from large-scale pre-trained datasets alleviates these issues, yet generic feature extraction suffers from domain shift, while indiscriminate fine-tuning risks over-fitting and elevated training budgets. To address the identified limitations, PlantClassiNet is implemented as a unified framework. This framework facilitates systematic comparative analysis of six CNN architectures, AlexNet, ResNet50, InceptionV3, MobileNetV3Small, DenseNet121 and EfficientNetB0, across three publicly available datasets: PlantVillage, PlantLeaves and Eggplant. Two alternative fine-tuning approaches are proposed: ‌Adaptive Fine-tuning (AdapFitu)‌, which adaptively determines the depth of unfrozen layers, learning rates, and reinitializes selected layers, and a ‌fixed-parameter baseline‌, which trains only the newly added classifier while keeping the convolutional backbone frozen and unfreezes a fixed number of network layers for retraining. Extensive experiments demonstrate that large models AlexNet, ResNet50, and Inceptionv3 achieve test accuracy exceeding 98.74% on the sizable PlantVillage dataset, whereas lightweight counterparts MobileNetV3Small, DenseNet121, and EfficientNetB0 achieve high accuracy of 99.79% ± 0.21% on the smaller Eggplant collection after fine-tuning. Full article

Fused deposition modeling of polyether ether ketone offers distinct advantages for fabricating complex and lightweight structures. Although three principal build orientations theoretically exist for practical 3D engineering components, research on their effects remains limited, especially regarding the influence of the interaction between build orientation and printing path on mechanical performance. This study investigated the tensile and shear properties, as well as the failure mechanisms, of FDM-fabricated PEEK under the coupled effects of build orientation and printing path through mechanical testing, fracture morphology analysis, and statistical methods. The results indicate that the printing path exerts a dominant influence on anisotropic behavior, while the interaction between printing path and build orientation jointly governs the shear failure modes. Under identical printing paths, the elongation at break varied by up to twofold across different build orientations, reaching a maximum of 96%, whereas samples printed with W or T paths exhibited elongations at break below 5%. Although shear and tensile moduli remained largely consistent across build orientations, other mechanical properties demonstrated significant differences. Variations in cross-sectional dimensions induced by build orientation markedly affected tensile performance: the coupled effect of build orientation and printing path was found to render the path repetition frequency a critical factor in determining temperature uniformity within the printed region and the quality of interlayer interfaces, thereby constituting the core mechanism underlying anisotropic behavior. Furthermore, larger cross-sections re-duced tensile modulus but enhanced yield strength and elongation at break, highlight-ing the regulatory role of cross-sectional geometry on mechanical response. Based on these findings, a synergistic optimization strategy integrating printing path, build orientation, and tensile–shear performance is proposed to achieve tailored mechanical properties in FDM-fabricated PEEK components. This approach enables controlled enhancement of structural performance to meet diverse application requirements. Full article

To address the challenges of varying scales of concrete defects, class imbalance, and hardware limitations, we propose EMAFG-RTDETR, a UAV-based concrete defect detection algorithm built upon RTDETR. In the feature extraction stage, a lightweight multi-scale attention feature extraction module (EMA-PRepFaster block) is designed, where PConv and RepConv are fused to improve the FasterNet block. At the same time, an Efficient Multi-scale Attention (EMA) module is introduced to enhance spatial feature extraction while reducing computational redundancy. For feature fusion, the Gather-and-Distribute mechanism of GOLD-YOLO is adopted to improve the fusion of multi-scale features. The introduction of Powerful-IoU v2 not only accelerates the training process but also enhances the model’s ability to capture defects of different sizes. To handle the issue of sample imbalance, a novel classification loss function, EMASVLoss, is proposed. This function adjusts classification loss values through piecewise weighting and integrates an exponential moving average mechanism for dynamic weight smoothing, improving model adaptability. Finally, the algorithm was deployed and validated on an octocopter UAV developed by our team. Experimental results demonstrate that EMAFG-RTDETR achieves a 2.5% improvement in mean Average Precision (mAP@0.5), reaching 90% on the concrete defect dataset, with reductions in both parameter size and computational cost. Moreover, the UAV equipped with the proposed algorithm can accurately detect cracks and spalling defects on concrete surfaces, validating the effectiveness of the improved model. Full article

Smoking in oilfield operation areas poses a severe risk of fire and explosion accidents, threatening production safety, workers’ lives, and the surrounding ecological environment. Such behavior represents a typical preventable unsafe human action. Detecting smoking behaviors among oilfield workers can fundamentally prevent such safety incidents. To address the challenges of low detection accuracy for small objects and frequent missed or false detections under extreme industrial environments, this paper proposes a GCEA-YOLO network based on YOLOv11 for smoking behavior detection. First, a CSP-EDLAN module is introduced to enhance fine-grained feature learning. Second, to reduce model complexity while preserving critical spatial information, an ADown module is incorporated. Third, an enhanced feature fusion module is integrated to achieve effective multiscale feature aggregation. Finally, an EfficientHead module is employed to generate high-precision and lightweight detection results. The experimental results demonstrate that, compared with YOLOv11n, GCEA-YOLO achieves improvements of 20.8% in precision, 6.9% in recall, and 15.1% in mean average precision (mAP). Overall, GCEA-YOLO significantly outperforms YOLOv11n. Full article

The physical layer security of Internet of Things (IoT) networks has become increasingly important but also introduces major security vulnerabilities due to the open and shared nature of wireless channels. Therefore, authentication and authorization remain critical challenges. To address these issues, this paper proposes a lightweight hybrid authentication framework that integrates Extended Kalman Filter (EKF)-based signal refinement with machine learning (ML) classification to strengthen device trust verification at the physical layer. The framework operates across device, edge, and cloud tiers, utilizing real-time received signal strength indicator (RSSI), link quality indicator (LQI), temperature, and battery level to generate unique device fingerprints. The EKF minimizes environmental noise and extracts stable signal characteristics, while the XGBoost classifier provides adaptive and efficient authentication. Experimental results show that the proposed hybrid model achieves 99.56% accuracy, a 99.71% F1-score, and a very low false acceptance rate. These findings confirm that the EKF–ML integration enhances signal stability and resistance to spoofing, offering a secure and scalable authentication solution for IoT networks. Full article

Auditability and regulatory compliance are increasingly required for deploying large language models (LLMs). Prior work typically targets isolated stages such as training or unlearning and lacks a unified mechanism for verifiable accountability across model updates. This paper presents AuditableLLM, a lightweight framework that decouples update execution from an audit-and-verification layer and records each update as a hash-chain-backed, tamper-evident audit trail. The framework supports parameter-efficient fine-tuning such as Low-Rank Adaptation (LoRA) and Quantized LoRA (QLoRA), full-parameter optimization, continual learning, and data unlearning, enabling third-party verification without access to model internals or raw logs. Experiments on LLaMA-family models with LoRA adapters and the MovieLens dataset show negligible utility degradation (below 0.2% in accuracy and macro-F1) with modest overhead (3.4 ms/step; 5.7% slowdown) and sub-second audit validation in the evaluated setting. Under a simple loss-based membership inference attack on the forget set, the audit layer does not increase membership leakage relative to the underlying unlearning algorithm. Overall, the results indicate that hash-chain-backed audit logging can be integrated into practical LLM adaptation, update, and unlearning workflows with low overhead and verifiable integrity. Full article

With the increasing prevalence of brain tumors, it becomes crucial to ensure fast and reliable segmentation in MRI scans. Medical professionals struggle with manual tumor segmentation due to its exhausting and time-consuming nature. Automated segmentation speeds up decision-making and diagnosis; however, achieving an optimal balance between accuracy and computational cost remains a significant challenge. In many cases, current methods trade speed for accuracy, or vice versa, consuming substantial computing power and making them difficult to use on devices with limited resources. To address this issue, we present PRA-UNet, a lightweight deep learning model optimized for fast and accurate 2D brain tumor segmentation. Using a single 2D input, the architecture processes four types of MRI scans (FLAIR, T1, T1c, and T2). The encoder uses inverted residual blocks and bottleneck residual blocks to capture features at different scales effectively. The Convolutional Block Attention Module (CBAM) and the Spatial Attention Module (SAM) improve the bridge and skip connections by refining feature maps and making it easier to detect and localize brain tumors. The decoder uses depthwise separable convolutions, which significantly reduce computational costs without degrading accuracy. The BraTS2020 dataset shows that PRA-UNet achieves a Dice score of 95.71%, an accuracy of 99.61%, and a processing speed of 60 ms per image, enabling real-time analysis. PRA-UNet outperforms other models in segmentation while requiring less computing power, suggesting it could be suitable for deployment on lightweight edge devices in clinical settings. Its speed and reliability enable radiologists to diagnose tumors quickly and accurately, enhancing practical medical applications. Full article

Cracks serve as a critical indicator of tunnel structural degradation. Manual inspections are difficult to meet engineering requirements due to their time-consuming and labor-intensive nature, high subjectivity, and significant error rates, while traditional image processing methods exhibit poor performance under complex backgrounds and irregular crack morphologies. To address these limitations, this study developed a high-quality dataset of tunnel crack images and proposed an improved lightweight semantic segmentation network, LiteSqueezeSeg, to enable precise crack identification and quantification. The model was systematically trained and optimized using a dataset comprising 10,000 high-resolution images. Experimental results demonstrate that the proposed model achieves an overall accuracy of 95.15% in crack detection. Validation on real-world tunnel surface images indicates that the method effectively suppresses background noise interference and enables high-precision quantification of crack length, average width, and maximum width, with all relative errors maintained within 5%. Furthermore, an integrated intelligent detection system was developed based on the MATLAB (R2023b) platform, facilitating automated crack feature extraction and standardized defect grading. This system supports routine tunnel maintenance and safety assessment, substantially enhancing both inspection efficiency and evaluation accuracy. Through synergistic innovations in lightweight network architecture, accurate quantitative analysis, and standardized assessment protocols, this research establishes a comprehensive technical framework for tunnel crack detection and structural health evaluation, offering an efficient and reliable intelligent solution for tunnel condition monitoring. Full article

This paper presents an experimental determination of the long-term mechanical properties of lightweight concrete with sintered aggregate under cyclic loading and the corresponding analytical standard models. The research was designed around two concrete mixtures. Multiple tests were conducted at the Building Structures, Geotechnics and Concrete Laboratory of the Building Research Institute (ITB), using various equipment including creep-testing machines and tensometric measurements of sample deformations. As a result of these tests, in addition to strength properties, the following time-dependent parameters were determined: the secant modulus of elasticity, shrinkage strains, and creep-recovery strains under cyclic loading. For the parameterization and modeling of constitutive equations, an analysis of creep strains under cyclic loads was carried out, taking into account the integral hereditary law according to the Boltzmann superposition principle and the long-term models formulated according to the following standards and pre-standards: Eurocode 2 (2004), Model Code 2010, Model Code 2020, and Eurocode 2 (2023). The results from the individual models were compared with the test results using the rules for evaluating correction factors for models determined according to Eurocode 2 (2023). It was concluded that the development of creep strain is correctly modeled by the aforementioned standard methods, albeit with the aforementioned correction factors. One of the research objectives was to determine whether the ratchetting phenomenon could be observed during creep of the tested concrete under cyclic loading; however, due to the very low level of plastic deformation, this phenomenon was not detected. The research confirmed the suitability of lightweight concrete with sintered aggregate for use in cyclically loaded concrete structures. Full article

Synthetic Aperture Radar (SAR) target detection faces significant challenges including speckle noise interference, weak small object features, and multi-category imbalance. To address these issues, this paper proposes LSDA-YOLO, an enhanced SAR target detection framework built upon the YOLO architecture that integrates Large Kernel Attention and SimAM dual attention mechanisms. Our method effectively overcomes these challenges by synergistically combining global context modeling and local detail enhancement to improve robustness and accuracy. Notably, this framework leverages the inherent symmetry properties of typical SAR targets (e.g., geometric symmetry of ships and bridges) to strengthen feature consistency, thereby reducing interference from asymmetric background clutter. By replacing the baseline C2PSA module with Deformable Large Kernel Attention and incorporating parameter-free SimAM attention throughout the detection network, our approach achieves improved detection accuracy while maintaining computational efficiency. The deformable large kernel attention module expands the receptive field through synergistic integration of deformable and dilated convolutions, enhancing geometric modeling for complex-shaped targets. Simultaneously, the SimAM attention mechanism enables adaptive feature enhancement across channel and spatial dimensions based on visual neuroscience principles, effectively improving discriminability for small targets in noisy SAR environments. Experimental results on the RSAR dataset demonstrate that LSDA-YOLO achieves 80.8% mAP50, 53.2% mAP50-95, and 77.6% F1 score, with computational complexity of 7.3 GFLOPS, showing significant improvement over baseline models and other attention variants while maintaining lightweight characteristics suitable for real-time applications. Full article

Accurate user positioning in 6G networks is essential for next-generation mobile services. However, classical approaches such as time-difference-of-arrival (TDoA) remain vulnerable to dense multipath and NLoS conditions commonly found in indoor and industrial environments. This paper proposes an AI-driven RF fingerprinting framework that leverages uplink channel state information (CSI) to achieve robust and privacy-preserving 2D localization. A lightweight convolutional neural network (CNN) extracts location-specific spectral–spatial fingerprints from CSI tensors, while a federated learning (FL) scheme enables distributed training across multiple gNBs without sharing raw channel data. The proposed integration of CSI tensor processing with FL and structured pruning is introduced as a novel solution for practical 6G edge positioning. To further reduce latency and communication costs, a structured pruning mechanism compresses the model by 40–60%, lowering the memory footprint with negligible accuracy loss. A performance evaluation in 3GPP-compliant indoor factory scenarios indicates a median positioning error below 1 m for over 90% of cases, significantly outperforming TDoA. Moreover, the compressed FL model reduces the FL communication load by ~38% and accelerates local training, establishing an efficient, secure, and deployment-ready positioning solution for 6G networks. Full article

Modern deep neural network (DNN) accelerators must sustain high throughput while avoiding performance degradation from supply voltage (VDD) droop, which occurs when large arrays of multiply–accumulate (MAC) units switch concurrently and induce high peak current ($IC{C}_{max}$) transients on the power delivery network (PDN). In this work, we focus on ASIC-class DNN accelerators with tightly synchronized MAC arrays rather than FPGA-based implementations, where such cycle-aligned switching is most pronounced. Conventional guardbanding and reactive countermeasures (e.g., throttling, clock stretching, or emergency DVFS) either waste energy or incur non-trivial throughput penalties. We propose SparseDroop, a unified hardware-conscious framework that proactively shapes instantaneous current demand to mitigate droop without reducing sustained computing rate. SparseDroop comprises two complementary techniques. (1) SparseStagger, a lightweight hardware-friendly droop scheduler that exploits the inherent unstructured sparsity already present in the weights and activations—it does not introduce any additional sparsification. SparseStagger dynamically inspects the zero patterns mapped to each processing element (PE) column and staggers MAC start times within a column so that high-activity bursts are temporally interleaved. This fine-grain reordering smooths $ICC$ trajectories, lowers the probability and depth of transient VDD dips, and preserves cycle-level alignment at tile/row boundaries—thereby maintaining no throughput loss and negligible control overhead. (2) SparseBlock, an architecture-aware, block-wise-structured sparsity induction method that intentionally introduces additional sparsity aligned with the accelerator’s dataflow. By co-designing block layout with the dataflow, SparseBlock reduces the likelihood that all PEs in a column become simultaneously active, directly constraining $IC{C}_{max}$ and peak dynamic power on the PDN. Together, SparseStagger’s opportunistic staggering (from existing unstructured weight zeros) and SparseBlock’s structured, layout-aware sparsity induction (added to prevent peak-power excursions) deliver a scalable, low-overhead solution that improves voltage stability, energy efficiency, and robustness, integrates cleanly with the accelerator dataflow, and preserves model accuracy with modest retraining or fine-tuning. Full article

Modern cryptographic systems increasingly depend on certified hardware modules to guarantee trustworthy key management, tamper resistance, and secure execution across Internet of Things (IoT), embedded, and cloud infrastructures. Although numerous FIPS 140-certified platforms exist, prior studies typically evaluate these solutions in isolation, offering limited insight into their cross-domain suitability and practical deployment trade-offs. This work addresses this gap by proposing a unified, multi-criteria evaluation framework aligned with the FIPS 140 standard family (including both FIPS 140-2 and FIPS 140-3), replacing the earlier formulation that assumed an exclusive FIPS 140-3 evaluation model. The framework systematically compares secure elements (SEs), Trusted Platform Modules (TPMs), embedded Systems-on-Chip (SoCs) with dedicated security coprocessors, enterprise-grade Hardware Security Modules (HSMs), and cloud-based trusted execution environments. It integrates certification analysis, performance normalization, physical-security assessment, integration complexity, and total cost of ownership. Validation is performed using verified CMVP certification records and harmonized performance benchmarks derived from publicly available FIPS datasets. The results reveal pronounced architectural trade-offs: lightweight SEs offer cost-efficient protection for large-scale IoT deployments, while enterprise HSMs and cloud enclaves provide high throughput and Level 3 assurance at the expense of increased operational and integration complexity. Quantitative comparison further shows that secure elements reduce active power consumption by approximately 80–85% compared to TPM 2.0 modules (<20 mW vs. 100–150 mW) but typically require 2–3× higher firmware-integration effort due to middleware dependencies. Likewise, SE050-based architectures deliver roughly 5× higher cryptographic throughput than TPMs (∼500 ops/s vs. ∼100 ops/s), whereas enterprise HSMs outperform all embedded platforms by two orders of magnitude (>10 000 ops/s). Because the evaluated platforms span both FIPS 140-2 and FIPS 140-3 certifications, the comparative analysis interprets their security guarantees in terms of requirements shared across the FIPS 140 standard family, rather than attributing all properties to FIPS 140-3 alone. No single architecture emerges as universally optimal; rather, platform suitability depends on the desired balance between assurance level, scalability, performance, and deployment constraints. The findings offer actionable guidance for engineers and system architects selecting FIPS-validated hardware for secure and compliant digital infrastructures. Full article

Lightweight vision models are widely deployed in mobile and embedded systems, where strict computational and memory budgets demand compact architectures. However, their evaluation remains dominated by ImageNet—a single, large natural-image dataset that requires substantial training resources. This creates a dilemma: lightweight models trained on ImageNet often reach capacity limits due to their constrained size, while scaling them to billions of parameters with specialized training tricks to achieve top-tier ImageNet accuracy does not guarantee proportional performance once the architectures are scaled back down to meet mobile constraints, particularly when re-evaluated on diverse data domains. These challenges raise two key questions: How should cross-dataset robustness be quantified in a simple and lightweight way, and which architectural elements consistently support generalization under tight resource constraints? To answer them, we introduce the Cross-Dataset Score (xScore), a simple metric that captures both average accuracy across domains and the stability of model rankings. Evaluating 11 representative lightweight models (2.5 M parameters) across seven datasets, we find that (1) ImageNet accuracy is a weak proxy for cross-domain performance, (2) xScore provides a simple and interpretable robustness metric, and (3) high-xScore models reveal architectural patterns linked to stronger generalization. Finally, the architectural insights and evaluation framework presented here provide practical guidance for measuring the xScore of future lightweight models. Full article

Three-phase AC motor drives play a key role in several applications, including energy conversion and automotive. Mainly, three-phase AC motor drives operate as closed loop control systems, where accurate feedback measurement sent by the current sensors is crucial to guarantee the good operation of the system. However, current sensors are potentially subject to several malfunctions that significantly affect the performance of the drive system. Accordingly, this paper proposes an efficient method for current sensor fault detection, and identification in three-phase AC motor drive system using a 2D Convolutional Neural Network (CNN). The proposed approach does need any additional extra-hardware components, since it uses only the signals already sent by the motor drive closed loop control. Indeed, it utilizes the 2D trajectory graph of the normalized motor current vector as input to a novel CNN Autoencoder model, which is introduced for feature extraction and classification. The efficiency and generalization capabilities of the proposed CNN autoencoder (PCAE) are benchmarked against a standard CNN model and conventional CNN autoencoders. The lightweight architecture of the PCAE enables its real-time implementation on a Raspberry pi 4 with a 750w experimental setup induction motor. The experimental results highlight that the proposed PCAE model can effectively detect and classify ten types of current sensor faults, in addition to distinguishing the healthy operation case. Moreover, the proposed approach achieves superior accuracy (99%), compared with conventional CNN (95%) and standard CNN-Autoencoder (96%) models. Full article