Search Results (110)

Plant diseases, being a subject of interdisciplinary research, significantly reduce crop yield, quality, and economic returns, while the misidentification of pathogens often leads to ineffective treatments and may harm beneficial organisms and ecosystems. This work develops an approach for robust visual classification of plant diseases under limited and heterogeneous data based on multi-scale fractal texture descriptors integrated into a convolutional neural network. The proposed method employs wavelet transform modulus maxima to extract two complementary fractal characteristics, local fractal dimension and singularity spectrum width, from leaf images at several spatial scales. These descriptors form multi-channel fractal maps fed into a fractal attention module (FAM) inserted after the third stage of a ResNet-50 architecture. The FAM learns to emphasize spatial regions where fractal properties are most discriminative, while a parallel branch encodes global fractal statistics into an auxiliary vector combined with backbone features at the final classification layer. Experiments are conducted on a large heterogeneous collection of 11 public plant disease datasets under 5-shot, 50-shot, and full-scale training regimes. The fractal-augmented model raises classification accuracy from 57.06% to 67.73% on 5 shots and from 80.81% to 86.11% on 50 shots, red outperforming the plain ResNet-50 in these settings, converges within 1–2 epochs versus 25–40, and shows markedly better resilience to color distortions, random occlusions, and grayscale conversion in most cases. The generated attention maps provide spatially explicit explanations of the model’s decisions, increasing transparency for practical use. The proposed approach demonstrates that fractal analysis, embedded as a modulating signal inside a deep network, can serve as an efficient and interpretable inductive bias, which is particularly valuable under data scarcity and noisy agricultural imagery. Full article

The cooperative tracking of multiple ground targets by multiple UAVs remains challenging under partial observability, limited communication, and obstacle constraints, owing to complex target association, difficult task handover, strong coupling among low-level continuous control decisions, and unstable critic value estimation. To address these issues, this paper proposes a hierarchical-guidance and Gaussian-process-corrected multi-agent proximal policy optimization method, termed HGP-MAPPO. Built upon the centralized-training and decentralized-execution paradigm, HGP-MAPPO introduces low-frequency task-guidance signals derived from target-association information, task handover and recovery cues, task priorities, and desired observation geometry. These guidance signals are incorporated as conditional inputs into the low-level actor–critic framework, thereby reducing the policy learning difficulty in jointly handling target tracking, occlusion recovery, obstacle avoidance, and smooth control. Moreover, to alleviate local estimation bias in the neural-network critic under complex partially observable conditions, a Gaussian-process-based residual correction mechanism is designed. Specifically, the posterior mean is used to compensate for value residuals, while the posterior uncertainty adaptively regulates the correction intensity, improving the stability of value evaluation and policy optimization. A sparse inducing-point approximation is adopted to control the training-stage computational cost, while the Gaussian-process module is removed during decentralized execution and, therefore, introduces no additional online inference overhead. Experiments are conducted in standard-obstacle and densely obstructed multi-UAV multi-target tracking scenarios, with DDPG-MHSA, MAPPO, MADDPG, and MATD3 adopted as baselines. The experimental results demonstrate that HGP-MAPPO achieves faster training convergence, higher average episode rewards, and improved target retention rates. It also effectively reduces UAV–target distance fluctuations and the mean absolute temporal-difference (TD) error. Ablation studies further confirm the contributions of task-guidance signals, Gaussian-process residual correction, and uncertainty-aware weighting to cooperative tracking performance and training stability. Full article

Wide- and narrow-band composite-axis servo systems in inter-satellite laser communication face critical challenges in balancing high-precision dynamic tracking and strong robust anti-disturbance performance under the coupling effect of high-speed inter-satellite relative motion and multiple strong disturbances. To address this issue, this paper proposes a composite control method integrating adaptive non-singular terminal sliding-mode control and a nonlinear extended state observer. First, a full-link dynamic model covering electromechanical coupling and inter-axis disturbance transmission is constructed to accurately quantify the disturbance characteristics of coarse- and fine-tracking loops. Second, a third-order nonlinear extended state observer is designed to realize real-time high-precision estimation and feedforward compensation of lumped disturbances. On this basis, a self-consistent adaptive non-singular terminal sliding-mode control law is formulated. Under the explicitly stated observer-residual and reaching-phase assumptions, the ideal continuous model provides finite-time convergence of the sliding variable and tracking error. Finally, a wide- and narrow-band cooperative strategy based on error frequency division is introduced to achieve complementary performance between large-stroke coarse tracking and ultra-high-precision fine tracking. Numerical simulations yield a steady-state tracking-error point estimate of 0.30 μrad and a 20 dB disturbance-suppression bandwidth of 1200 Hz. In the semi-physical dynamic-tracking test, the proposed controller limits the peak error to 1.2 μrad; the instrument-only expanded uncertainty of the detector output is estimated as 0.12 μrad (coverage factor k = 2). At the reported evaluation points, the proposed method outperforms PID, conventional sliding-mode control, and linear active-disturbance-rejection control. Deterministic robustness simulations also show smaller tracking errors and shorter recovery times under parameter perturbation, actuator saturation, and temporary link occlusion. No Monte Carlo loss-of-lock probability is claimed. Full article

Manual fruit thinning is labor-intensive and inefficient, making the development of intelligent visual detection systems a crucial approach for improving the quality and production efficiency of the pear industry. However, in natural orchard environments, young pear fruits are small in size with slender fruit stalks, and their texture and color characteristics are highly similar to those of tender branches. Furthermore, variations in illumination and occlusions caused by branches and leaves make it difficult for existing detection models to simultaneously and accurately identify fruits and fruit stalks, limiting their application in automated thinning equipment. In this study, Yuluxiang pear was selected as the research object, and image data were collected under diverse field conditions, including forward-lighting, backlighting, close-range shooting, long-range shooting and fruit overlapping. A dedicated dataset containing 3057 images was established. Based on the RT-DETR-R18 network, a lightweight and high-precision fruit–stalk synchronous detection model was proposed. Specifically, the backbone network was reconstructed by integrating GCConv with C2f modules to enhance global feature extraction for slender fruit stalks. The bottleneck structure was optimized using GCConvC3 to reduce feature degradation under occlusion conditions, and an additional 4× down-sampling P2 detection head was introduced to improve the detection capability for small targets. To fully validate the model performance and stability, three types of experiments were conducted in this study: ablation experiments, repeated experiments with different random seeds, and comparative experiments. Ablation experiments verified the cumulative performance improvements brought by the introduced modules. Repeated experiments with different random seeds were performed to explore training randomness-induced performance fluctuations, and the results demonstrated that the proposed model maintains stable overall detection accuracy with minor metric fluctuations. Comparative experiments demonstrated that the proposed model achieved a compact parameter size of only 15.97 M, with a precision of 95.0% for young pear fruit detection and an mAP50 of 83.0% for fruit stalk detection, outperforming all comparative models in overall mAP50. The training convergence curves and Grad-CAM++ visualization results further confirmed the stable optimization process and enhanced feature attention capability of the proposed model. By achieving a favorable balance between detection accuracy and model lightweightness, this approach provides effective technical support for the development of intelligent fruit thinning equipment and vision-based systems for smart pear orchards. Full article

Rock bolts are the primary ground support mechanism in underground mining. Yet verification of their installation is rarely captured in a spatially precise, retrievable form, leaving operators without an auditable as-built record for regulatory review or post-incident reconstruction. This paper presents an automated rock bolt detection process that closes this documentation gap using dense point clouds from an underground hard rock mine acquired by terrestrial laser scan. The method computes per-point ambient occlusion (AO) on closure plane-sealed chambers using a PCV implementation of the ShadeVIS principle, forms candidates from a multi-scale protrusion field, and segments them by prominence watershed before classifying each candidate with PCA-based geometric descriptors, without machine learning or training data. Installation perpendicularity is applied as a per-detection confidence cue, and detections are reported in confidence tiers that concentrate human review on the ambiguous minority. Validated against a database of 1447 bolts across 20 walls in two areas of an underground mine, the system achieved an overall recall of 83.7%, with human review completing the inventory to 100%. The physics-based design transfers across bolt types and mine geometries through parameter re-tuning rather than retraining, addressing the core limitation of deep learning methods, which require site-specific labelled datasets. Full article

The perception abilities of autonomous vehicles are highly dependent on the configuration of various sensors installed in the vehicle. However, sensor placement often comes from a manual process. This research paper introduces an optimization model to address sensor placement as a multi-objective problem where the objectives consist of the maximization of weighted coverage of Regions of Interest (ROIs) and the minimization of sensor cost, subject to physical and perceptual constraints such as mounting bounds, directional balance, redundancy, and self-occlusion. The introduced method combines NSGA-II algorithm with visibility analysis by means of ray casting and weighted coverage of ROIs near the vehicle. The developed framework has been tested on three different vehicle types: a small car, a light commercial transporter, and a large bus. It has been found that the suggested optimization technique outperforms the baseline solution in case of the car and the bus, whereas the transporters pose much harder optimization problems characterized by high variance between runs. Thus, the results show that the sensor suite design heavily depends on the platform type and geometrical symmetry hinders the convergence of an evolutionary search method. Full article

Profiling side-channel attacks based on deep learning can recover cryptographic information from power-consumption traces, but their effectiveness may be reduced when desynchronisation displaces informative leakage across temporal positions. This study investigates whether Temporal Convolutional Networks (TCNs) can model the local and long-range dependencies present in desynchronised traces. Three complementary TCN variants are designed to isolate different temporal modelling strategies: a single-kernel dilated architecture (TCN1), a multi-scale architecture using parallel kernel sizes (TCN2), and a residual architecture intended to support stable hierarchical feature learning (TCN3). The models are evaluated on the ASCAD v1 fixed-key benchmark under synchronised conditions and maximum temporal shifts of 25, 50, and 75 samples, using validation loss and complementary key-ranking metrics. Under the most challenging setting, TCN1 achieves the highest Rank Success Rate and reaches its best rank substantially earlier than the conventional CNN baselines, although the lowest Final Rank is obtained by a CNN. These results indicate that TCN1 provides the most favourable trade-off among convergence speed, ranking consistency, and architectural complexity, without establishing uniform TCN superiority across all metrics. Gradient saliency, LIME, and occlusion analyses further identify temporal regions that influence the model predictions and are consistent with expected leakage patterns. The main contribution is a controlled comparison of complementary TCN design strategies, combined with an explainability analysis, for profiling side-channel attacks under trace desynchronisation. Full article

Objective: To determine and compare the center of resistance (Cres) for maxillary protraction in fused and patent palatal suture models using three-dimensional finite element analysis. Study design: A three-dimensional craniofacial model was constructed from CBCT data of a 10-year-old female with maxillary hypoplasia. Models with fused and patent palatal sutures were created. Finite element analysis evaluated displacement patterns following anteroposterior forces (10 N) at different vertical positions (H1–H4) and superoinferior forces (10 N) at different sagittal positions (V1–V5). The force application site producing the most uniform displacement with minimal rotation was identified as the Cres. Results: In the fused palatal suture model, the Cres was located at H3.3 (infraorbital rim level, 15 mm superior to the line midway between the occlusal plane and the infraorbital rim.) for anteroposterior loading and at V3.6 (slightly distal to the lower border of the zygomatic process, 15 mm anterior to the posterior nasal spine) for superoinferior loading. In the patent palatal suture model, the Cres remained at similar levels but shifted laterally, requiring force application 20 mm lateral to the midpalatal suture to achieve translational movement. Force application closer to the midpalatal suture caused the maxillary segments to diverge anteriorly and converge posteriorly, whereas force application at the lateral maxillary rim produced the opposite rotational pattern. Conclusion: The Cres of the nasomaxillary complex was consistently located near the infraorbital rim under anteroposterior loading and slightly distal to the zygomatic process under superoinferior loading in both suture configurations. In patent sutures, optimal translational movement required bilateral force application 20 mm lateral to the midpalatal suture. These findings provide subject-specific biomechanical information that may assist in the design and evaluation of maxillary protraction force systems. Full article

Background & Objectives: Endovascular embolization has matured into a sophisticated, precision-guided discipline that is central to the management of complex neurovascular pathologies. This review synthesizes contemporary treatment strategies, evaluating the advanced material characteristics of conventional inert liquid polymers, specifically non-adhesive ethylene vinyl alcohol (EVOH) copolymers and adhesive cyanoacrylates, alongside their targeted clinical applications in brain arteriovenous malformations (bAVMs), dural arteriovenous fistulas (dAVFs), hypervascular intracranial tumors, and chronic subdural hematomas (CSDHs). Furthermore, it examines the critical material and hemodynamic constraints that limit these agents in cerebral aneurysm repair. Methods: A comprehensive literature synthesis through 3 July 2026 was integrated with peer-reviewed clinical illustrations to evaluate both procedural mechanics and the necessity of post-procedural physiological management. Review Findings: Embolization serves a critical dual role: as a definitive curative therapy and as an essential preoperative or radiosurgical adjunct. As demonstrated by recent clinical validations, technical angiographic success must be closely coupled with vigilant neurocritical oversight to manage profound, localized hemodynamic shifts. While these conventional methods represent established clinical practice, the field is evolving away from inert mechanical occlusion toward a highly integrated approach. The convergence of stimuli-responsive “smart” hydrogels and endovascular robotics is being evaluated for potential roles in transforming these interventions into dynamic, bioactive platforms capable of modulating disease-specific mechanisms, such as Rat Sarcoma-Mitogen-Activated Protein Kinase (RAS-MAPK) and Bone Morphogenetic Protein (BMP) signaling in bAVMs or the Von Hippel-Lindau/Vascular Endothelial Growth Factor (VHL/VEGF) axis in hypervascular tumors. This review further analyzes landmark data, including the Squid Trial For the Embolization of the Middle Meningeal Artery for Treatment of Chronic Subdural Hematoma (STEM) trial for CSDH, providing a synthesis for translating these advanced material sciences into standardized, multidisciplinary neurointerventional care. Full article

Accurate boxing action recognition is critical for intelligent combat training, action quality assessment, and sports injury prevention. However, existing deep learning approaches face three key challenges: limited feature extraction for high-speed non-rigid boxing motions, weak robustness against background interference and occlusion, and performance instability from labor-intensive manual hyperparameter tuning. Furthermore, the original Centered Collision Optimizer (CCO), a biomimetic algorithm inspired by celestial collision dynamics, suffers from insufficient population diversity, poor adaptive regulation, and premature convergence in high-dimensional hyperparameter optimization tasks. To address these issues, this paper proposes a novel biomimetic optimization-driven boxing action recognition framework, where an Enhanced Adaptive Centered Collision Optimizer (EACCO) automatically optimizes the hyperparameters of a Multi-Scale Spatio-Temporal Adaptive ConvNeXt (MSTA-ConvNeXt) network. First, the MSTA-ConvNeXt backbone integrates multi-scale dynamic deformable convolution, a Bi-GRU spatio-temporal fusion module, and a dual-channel attention mechanism to enhance fine-grained feature extraction and temporal modeling. Second, three biomimetic improvements are introduced to CCO: Tent chaotic elite opposition-based initialization, adaptive nonlinear convergence factor with dynamic weight guidance, and adaptive Gaussian-Cauchy hybrid mutation, which balance exploration and exploitation and avoid local optima. Experiments on two public benchmark datasets show that the proposed framework achieves 96.1% accuracy, 95.9% precision, 95.7% recall, and 95.8% F1-score on the Boxing Jab Skeleton Dataset, and 95.4% accuracy, 95.2% precision, 94.9% recall, and 95.0% F1-score on the Olympic Boxing dataset, outperforming all state-of-the-art methods. Ablation studies validate the effectiveness of each EACCO component and confirm that this biomimetic hyperparameter optimization approach outperforms manual tuning and other popular optimizers. This work provides an effective biomimetic optimization solution for intelligent sports action recognition. 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

Reliable visual detection of small floating objects on the water surface is a prerequisite for environmental monitoring and clean-up tasks performed by unmanned surface vehicles (USVs) on inland waterways. Such scenes are routinely degraded by low illumination at dawn and dusk, strong specular reflections, ripple-induced clutter, and large object-scale variations, which together cause missed detections, false alarms, and unstable localization. Aiming at these practical challenges, this study conducts a scenario-oriented optimization and experimental validation based on the lightweight YOLOv11n detector. We integrate multiple mature attention mechanisms, regression loss functions and data augmentation strategies to develop an improved scheme, YOLOv11-LREP, for floating object detection. The detailed optimizations are as follows: (i) a Coordinate Attention (CoordAtt) module is inserted at the top of the backbone to enhance positional encoding and highlight obstacle-related semantic regions; (ii) three Efficient Channel Attention (ECA) modules are embedded at the multi-scale fusion nodes of the Neck so that reflection- and ripple-induced spurious channel responses can be suppressed at almost no extra cost; (iii) the Powerful-IoU (PIoU) loss replaces the original regression loss to enforce four-side boundary alignment and stabilize convergence on small, blurred-edge targets; and (iv) a joint low-light and reflection augmentation strategy, together with CutMix region-level mixing, broadens the training distribution along the illumination and occlusion axes. Experiments on the public FloW-Img dataset, split into 1200 training and 800 validation images (2024 instances) and run under a fixed random seed (seed = 0, deterministic = true), show that YOLOv11-LREP attains AP₅₀ = 80.1%, AP_50:95 = 38.5%, and AP_S = 24.3% with only 2.84 M parameters and 9.3 GFLOPs. On an NVIDIA RTX 4060 Laptop GPU, the model runs at 3.3 ms total per 640 × 640 image (≈303 FPS), satisfying real-time perception requirements while retaining lightweight deployability. The ablation results indicate that different components contribute differently to localization accuracy, small-object sensitivity, and robustness, and that the final configuration provides a balanced trade-off rather than the best value for every individual metric. A systematic threshold sensitivity analysis (F₁ fluctuation < 0.2%) demonstrates the stability of the final model. Full article

Infrared cameras used in radio telescopes often suffer image degradation in complex optical and thermal environments. Solar radiation, convergent reflected light, and thermal emission from support structures can substantially impair imaging performance. To address this problem, this paper proposes a split reflective ellipsoidal baffle for suppressing infrared imaging degradation. Unlike conventional baffles, which mainly rely on structural occlusion and surface absorption, the proposed design functions as an upstream stray light regulation unit. It also establishes a computational framework integrating ellipsoidal vane geometry, realistic edge microtopography modelling, ray-tracing simulation, and detector plane irradiance response analysis. First, the reflective properties of the ellipsoidal surface are used to construct an off-axis stray light propagation constraint model. Under this model, incident stray radiation is redirected away from the effective imaging path or guided into light-trapping regions between adjacent vanes. Second, a laser confocal microscope is used to capture the true three-dimensional edge morphology of vanes with different materials and machining angles. This strategy addresses the limitations of the conventional 0.02 mm rounded edge approximation, which cannot accurately represent real scattering behaviour. The measured morphologies are then converted into high-fidelity computational models compatible with ray-tracing analysis. Furthermore, stray light suppression performance is evaluated using point source transmittance, detector plane irradiance distribution, and grey scale response in experimental images. Simulation and darkroom experiments show that the proposed baffle suppresses residual stray light more effectively than conventional absorptive baffles. The results demonstrate a computable, manufacturable, and experimentally verifiable strategy for front-end stray light control and baffle optimisation. This strategy can also support image quality enhancement in infrared imaging systems operating under complex optical and thermal environments. Full article

Dynamic hospital corridor environments are characterized by complex corridor environments, diverse target-scale variations, frequent occlusions, and dense small-object distribution, posing significant challenges to the accuracy and efficiency of the existing methods on resource-constrained platforms. To effectively address these challenges, a high-precision framework CSAWOA (Cross Search Adaptive Whale Optimization Algorithm)-YOLOv8 (You Only Look Once version 8) model for complex medical environments was introduced in this work. By jointly modelling high-level semantic information and low-level cues such as texture and colour, the proposed model achieved a more discriminative and informative feature representation. The T-CBS (Transformer-Convolutional Bottleneck Structure) module, capable of extracting shallow-level features and integrating global contextual information to address target occlusion issues, was also proposed. Furthermore, the integration of the BiFormer module yielded an enhanced feature discriminability, improving small-target recognition while reducing sensitivity to background noise. The classification function was modified, effectively solving the problem of class imbalance in complex corridor environments. The combination of these two concepts achieved an effective balance of diversity in detection and convergence speed, leading to improved optimization performance and greater resistance to local-optimum stagnation. Meanwhile, an improved version of the WOA was developed, termed CSAWOA, enabling automatic hyperparameter optimization for the improved YOLOv8 model. From the experimental results, improvements of 4.9%, 6.1%, and 8.3% in mAP, precision, and recall, respectively, compared to YOLOv8 were demonstrated, while also exhibiting better generalization. Overall, the proposed method provides a reliable and efficient approach for object detection in complex hospital corridors, offering a valuable foundation for future research and real-world healthcare applications. Full article

Background: Ischemic stroke remains a leading cause of mortality and disability worldwide. Despite extensive preclinical research, most neuroprotective strategies have failed to translate into clinical benefit, partly due to methodological variability. The transient intraluminal filament middle cerebral artery occlusion (tifMCAO) model is widely used, yet its implementation lacks consistency. This review aimed to characterize tifMCAO methodologies in adult rats and examine how experimental variability relates to reported outcomes. Methods: A systematic review was conducted following PRISMA guidelines. Studies using tifMCAO in adult rats were included. MEDLINE (PubMed), Web of Science, and Scopus were searched up to March 2025. Risk of bias was assessed using the SYRCLE tool and reporting quality using the ARRIVE checklist. The protocol was registered in PROSPERO (CRD420251140869). Results were synthesized narratively. Results: A total of 125 studies were included. A commonly used framework involved male Sprague–Dawley rats (6–12 weeks), silicone-coated monofilaments, occlusion durations of 60–120 min (most frequently 90 min), and isoflurane anesthesia, although this reflects methodological convergence rather than true standardization. Substantial variability was observed across methodological parameters. Variations in ischemia duration, filament properties, and anesthesia were associated with differences in infarct size, blood–brain barrier disruption, and functional outcomes. Conclusions: The tifMCAO model shows partial methodological convergence alongside significant variability influencing outcomes. Improved standardization and reporting are essential to enhance reproducibility and translational relevance. Full article