Search Results (1,373)

Most Transformer-based tracking methods overemphasize tracking accuracy while neglecting tracking efficiency. To address this issue, a lightweight object tracking method based on a hierarchical pruning strategy is proposed. First, an activation module is introduced to adaptively adjust the attention layers of the backbone network, avoiding unnecessary computations of attention layers. Then, based on the demands at different stages of the network, the standard attention mechanism is improved and divided into hybrid attention and cross attention, further reducing redundant computations. Experimental results on multiple benchmark datasets demonstrate that PSTrack achieves a tracking speed of 216 FPS on GPU and 48 FPS on CPU while maintaining competitive accuracy, with an AO of 68.7% on GOT-10k and an AUC of 66.1% on LaSOT. The proposed tracking method exhibits strong competitiveness in both quantitative and qualitative evaluations. Full article

Effective and transparent medical diagnosis relies on accurate and interpretable classification of medical images across multiple modalities. This paper introduces an explainable multi-modal image analysis framework based on a dual-stream architecture that fuses handcrafted descriptors with deep features extracted from a custom MobileNet. Handcrafted descriptors include frequency-domain and texture features, while deep features are summarized using 26 statistical metrics to enhance interpretability. In the fusion stage, complementary features are combined at both the feature and decision levels. Decision-level integration combines calibrated soft voting, weighted voting, and stacking ensembles with optimized classifiers, including decision trees, random forests, gradient boosting, and logistic regression. To further refine performance, a hybrid class-specific feature selection strategy is proposed, combining mutual information, recursive elimination, and random forest importance to select the most discriminative features for each class. This hybrid selection approach eliminates redundancy, improves computational efficiency, and ensures robust classification. Explainability is provided through Local Interpretable Model-Agnostic Explanations, which offer transparent details about the ensemble model’s predictions and link influential handcrafted features to clinically meaningful image characteristics. The framework is validated on three benchmark datasets, i.e., BTTypes (brain MRI), Ultrasound Breast Images, and ACRIMA Retinal Fundus Images, demonstrating generalizability across modalities (MRI, ultrasound, retinal fundus) and disease categories (brain tumor, breast cancer, glaucoma). Full article

Feed formulation is a typical multi-objective optimization problem that aims to minimize cost while satisfying multiple nutritional constraints. However, existing methods often suffer from limitations in handling nonlinear constraints, high-dimensional decision spaces, and solution feasibility. To address these challenges, this study proposes a multi-objective feed formulation method based on an improved Non-dominated Sorting Genetic Algorithm II (NSGA-II). A hybrid Dirichlet–Latin Hypercube Sampling (Dirichlet-LHS) strategy is introduced to generate an initial population with high feasibility and diversity, together with an iterative normalization-based dynamic repair operator to efficiently handle ingredient proportion and nutritional constraints. In addition, an adaptive termination mechanism based on the hypervolume improvement rate (Hypervolume Termination, HVT) is designed to avoid redundant computation while ensuring effective convergence of the Pareto front. Experimental results demonstrate that the Dirichlet–LHS strategy outperforms random sampling, Dirichlet sampling, and Latin hypercube sampling in terms of hypervolume and solution diversity. Under identical nutritional constraints, the improved NSGA-II reduces formulation cost by 1.52% compared with multi-objective Bayesian optimization and by 2.17% relative to conventional feed formulation methods. In a practical application to meat sheep diet formulation, the optimized feed cost is reduced to 1162.23 CNY per ton, achieving a 4.83% cost reduction with only a 1.09 s increase in computation time. These results indicate that the proposed method effectively addresses strongly constrained multi-objective feed formulation problems and provides reliable technical support for precision feeding in intelligent livestock production. Full article

To achieve accurate voltage sag prediction and early warning, thereby improving power quality, a hybrid voltage sag prediction framework is proposed by integrating Kernel Entropy Component Analysis (KECA) with an improved Northern Goshawk Optimization (NGO) algorithm for hyperparameter tuning of a CNN-LSTM model. First, to address the limitations of the original NGO, such as proneness to falling into local optima and high randomness of the initial population distribution, a refraction-opposition-based learning mechanism is introduced to enhance population diversity and expand the search space. Furthermore, a sine–cosine strategy (SCA) with nonlinear weight coefficients is integrated into the exploration phase to dynamically adjust the search step size, optimizing the balance between global exploration and local exploitation, thereby boosting convergence speed and accuracy. The improved algorithm (SCNGO) is then utilized to optimize the hyperparameters of the CNN-LSTM model. Second, KECA is applied to voltage-sag-related data to extract key features and eliminate redundant information, and the resulting dimensionally reduced data are fed as input to the SCNGO-CNN-LSTM model to further improve prediction performance. Experimental results demonstrate that the SCNGO-CNN-LSTM model outperforms other comparative models significantly across multiple evaluation metrics. Compared with NGO-CNN-LSTM, GWO-CNN-LSTM, and the original CNN-LSTM, the proposed method achieves a mean squared error (MSE) reduction of 53.45%, 44.68%, and 66.76%, respectively. The corresponding root mean squared error (RMSE) is decreased by 25.33%, 18.61%, and 36.92%, while the mean absolute error (MAE) is reduced by 81.23%, 77.04%, and 86.06%, respectively. These results confirm that the proposed framework exhibits superior feature representation capability and significantly improves voltage sag prediction accuracy. Full article

To investigate the effects of different Trichosanthes kirilowii Maxim. varieties on leaf nutrients, soil nutrients, and microbial community composition, this study selected Yuelou No. 3 and Huiji No. 2, two major cultivars from the primary production area of Shishou City. The two varieties were cultivated at different locations under standardized agronomic management practices, and a systematic comparative analysis was carried out over a 10-month sampling period from March to December 2024. The analysis encompassed their leaf nutrients (total nitrogen, total phosphorus, total potassium, and relative chlorophyll content), soil nutrients (organic matter, alkali-hydrolyzable nitrogen, available phosphorus, and available potassium), and microbial community characteristics. The results revealed significant varietal differences in leaf nutrient content: the average total phosphorus content of Yuelou No. 3 (0.44%) was higher than that of Huiji No. 2 (0.39%), while Huiji No. 2 exhibited higher total nitrogen (3.73%), total potassium (3.86%), and SPAD (44.72). Leaf nutrient content in both varieties followed a pattern of nitrogen > potassium > phosphorus, with peak phosphorus and potassium demand occurring earlier in Yuelou No. 3. Additionally, Yuelou No. 3 contained higher organic matter (12.73 g/kg) and alkali-hydrolyzable nitrogen (103.02 mg/kg), while Huiji No. 2 showed enhanced soil pH (7.02), available phosphorus (6.96 mg/kg), and available potassium (180.00 mg/kg). Soil available nutrient dynamics displayed a pattern of slow change during the early stage, a rapid increase during the middle stage, and stabilization in the later stage. Microbial analysis revealed no significant differences in alpha diversity between the two varieties, although Yuelou No. 3 showed marginally higher diversity indices during early to mid-growth stages. In contrast, beta diversity showed significant separation in PCoA space. Proteobacteria, Acidobacteria, and Ascomycota were the dominant microbial phyla. Dominant genera included Kaistobacter, Mortierella, and Neocosmospora, among others, with variety-specific relative abundances. Redundancy analysis further supported the variety-specific influence of soil physicochemical properties on microbial community structure, with available phosphorus, available potassium, and alkali-hydrolyzable nitrogen identified as key factors shaping community composition. This study provides a theoretical basis for understanding the impact of different Trichosanthes kirilowii Maxim. varieties on soil–plant–microbe interactions and suggests potential directions for future research on fertilization and management strategies tailored to varietal differences. Full article

As an emerging immersive media format, point clouds (PC) inevitably suffer from distortions such as compression and noise, where even local degradations may severely impair perceived visual quality and user experience. It is therefore essential to accurately evaluate the perceived quality of PC. In this paper, a no-reference point cloud quality assessment (PCQA) method that uses complexity-driven patch sampling and an attention-enhanced Swin-Transformer is proposed to accurately assess the perceived quality of PC. Given that projected PC maps effectively capture distortions and that the quality-related information density varies significantly across local patches, a complexity-driven patch sampling strategy is proposed. By quantifying patch complexity, regions with higher information density are preferentially sampled to enhance subsequent quality-sensitive feature representation. Given that the indistinguishable response strengths between key and redundant channels during feature extraction may dilute effective features, an Attention-Enhanced Swin-Transformer is proposed to adaptively reweight critical channels, thereby improving feature extraction performance. Given that traditional regression heads typically use a single-layer linear mapping, which overlooks the heterogeneous importance of information across channels, a gated regression head is designed to enable adaptive fusion of global and statistical features via a statistics-guided gating mechanism. Experiments on the SJTU-PCQA dataset demonstrate that the proposed method consistently outperforms representative PCQA methods. Full article

Salvia miltiorrhiza Bunge is a traditional Chinese medicinal plant whose roots are rich in water-soluble phenolic acids. Rosmarinic acid and salvianolic acid B are representative components that confer antibacterial, antioxidant, and cardio-cerebrovascular protective activities. However, these metabolites often accumulate at low and unstable levels in planta, which limits their efficient development and use. This review summarises recent advances in understanding salvianolic acid biosynthesis and its transcriptional regulation in S. miltiorrhiza. Current evidence supports a coordinated pathway composed of the phenylpropanoid route and a tyrosine-derived branch, which converge to generate rosmarinic acid and subsequently more complex derivatives through oxidative coupling reactions. Key findings on transcription factor families that fine-tune pathway flux by regulating core structural genes are synthesised. Representative positive regulators such as SmMYB111, SmMYC2, and SmTGA2 activate key nodes (e.g., PAL, TAT/HPPR, RAS, and CYP98A14) to promote phenolic acid accumulation. Conversely, negative regulators such as SmMYB4 and SmMYB39 repress pathway genes and/or interfere with activator complexes. Major regulatory features include hormone-inducible signalling, cooperative regulation through transcription factor complexes, and emerging post-transcriptional and post-translational controls. Future directions and challenges are discussed, including overcoming regulatory redundancy and strong spatiotemporal specificity of transcriptional control. Integrating spatial and single-cell omics with functional genomics (e.g., genome editing and rational TF stacking) is highlighted as a promising strategy to enable predictive metabolic engineering for the stable, high-yield production of salvianolic acid-type compounds. Full article

This study investigates Armenian editorial conventions for inscriptions and evaluates their compatibility and the possibility of their further integration with international standards of epigraphic editing for open access and equal use. It focuses on the Divan Hay Vimagrut’yan (Corpus of Armenian Epigraphy), launched in the 1960s, which introduced a systematic apparatus for distinguishing diplomatic transcriptions from interpretative reconstructions. Later Armenian publications often simplified these conventions, replacing specialized signs with typographic substitutes. While these changes improved accessibility, they also reduced palaeographic precision and created inconsistencies across editions. Through comparative analysis with the Leiden Conventions and the EpiDoc TEI framework, the research identifies both areas of alignment and points of divergence. Armenian conventions handle missing letters, restorations, redundancies, and abbreviations in distinctive ways, sometimes reassigning the meaning of symbols across different publications. This variation, if not explicitly documented, complicates digital encoding and risks loss of information. Methodologically, this study develops a digital heritage interoperability model that translates local Armenian editorial practices into machine-actionable standards, enabling their integration into international infrastructures such as EpiDoc and FAIR-based cultural heritage systems. The principal contribution of this work is the proposal of a dual-track encoding strategy. One track applies a granular mapping of Armenian signs to the full set of Leiden and EpiDoc categories, ensuring maximum interoperability. The other track preserves a simplified schema faithful to Armenian usage, reflecting local scholarly traditions. Together, these approaches provide both international comparability and cultural specificity. The conclusion is that Armenian inscriptions can be effectively integrated into global digital infrastructures by means of transparent documentation, crosswalk tables, and encoding policies that follow FAIR principles. This ensures long-term preservation, machine-actionability, and the broader reuse of Armenian epigraphic data in comparative cultural heritage research. Full article

Robotic peach harvesting represents a pivotal strategy for reducing labor costs and improving production efficiency. The fundamental prerequisite for a harvesting robot to successfully complete picking tasks is the accurate recognition of fruit growth posture subsequent to target identification. This study proposes a novel methodology for peach growth posture recognition by integrating an enhanced YOLOv8 algorithm with the RTMpose keypoint detection framework. Specifically, the conventional Neck network in YOLOv8 was replaced by an Atrous Feature Pyramid Network (AFPN) to bolster multi-scale feature representation. Additionally, the Soft Non-Maximum Suppression (Soft-NMS) algorithm was implemented to suppress redundant detections. The RTMpose model was further employed to locate critical morphological landmarks, including the stem and apex, to facilitate precise growth posture recognition. Experimental results indicated that the refined YOLOv8 model attained precision, recall, and mean average precision (mAP) of 98.62%, 96.3%, and 98.01%, respectively, surpassing the baseline model by 8.5%, 6.2%, and 3.0%. The overall accuracy for growth posture recognition achieved 89.60%. This integrated approach enables robust peach detection and reliable posture recognition, thereby providing actionable guidance for the end-effector of an autonomous harvesting robot. Full article

In this study, using global liner shipping schedules, UNCTAD’s Port Liner Shipping Connectivity Index and Liner Shipping Bilateral Connectivity Index, together with bilateral trade-value data for 2022–2024, we construct a multilayer weighted port-to-port network that explicitly embeds port-level cargo-handling and service organisation capabilities, as well as demand-side routing pressure, into node and edge weights. Building on this network, we apply CONCOR-based structural-equivalence analysis to delineate functionally homogeneous port clusters, and adopt a structural role identification framework that combines multi-indicator connectivity metrics with Rank-Sum Ratio–entropy weighting and Probit-based binning to classify ports into high-efficiency core, bridge-control, and free-form bridge roles, thereby tracing the reconfiguration of cluster-level functional structures before and after the Red Sea crisis. Empirically, the clustering identifies four persistent communities—the Intertropical Maritime Hub Corridor (IMHC), Pacific Rim Mega-Port Agglomeration (PRMPA), Southern Commodity Export Gateway (SCEG), and Euro-Asian Intermodal Chokepoints (EAIC)—and reveals a marked spatial and functional reorganisation between 2022 and 2024. IMHC expands from 96 to 113 ports and SCEG from 33 to 56, whereas EAIC contracts from 27 to 10 nodes as gateway functions are reallocated across clusters, and the combined share of bridge-control and free-form bridge ports increases from 9.6% to 15.5% of all nodes, demonstrating a thicker functional backbone under rerouting pressures. Spatially, IMHC extends from a Mediterranean-centred configuration into tropical, trans-equatorial routes; PRMPA consolidates its role as the densest trans-Pacific belt; SCEG evolves from a commodity-based export gateway into a cross-regional Southern Hemisphere hub; and EAIC reorients from an Atlantic-dominated structure towards Eurasian corridors and emerging bypass routes. Functionally, Singapore, Rotterdam, and Shanghai remain dominant high-efficiency cores, while several Mediterranean and Red Sea ports (e.g., Jeddah, Alexandria) lose centrality as East and Southeast Asian nodes gain prominence; bridge-control functions are increasingly taken up by European and East Asian hubs (e.g., Antwerp, Hamburg, Busan, Kobe), acting as secondary transshipment buffers; and free-form bridge ports such as Manila, Haiphong, and Genoa strengthen their roles as elastic connectors that enhance intra-cluster cohesion and provide redundancy for inter-cluster rerouting. Overall, these patterns show that resilience under the Red Sea crisis is expressed through the cluster-level rebalancing of core–control–bridge roles, suggesting that port managers should prioritise parallel gateways, short-sea and coastal buffers, and sea–land intermodality within clusters when designing capacity expansion, hinterland access, and rerouting strategies. Full article

Phishing attacks continue to evolve and exploit fundamental human impulses, such as trust and the need for a rapid response, as well as emotional triggers. This makes the human mind both a valuable asset and a significant vulnerability. The proliferation of zero-day vulnerabilities has been identified as a significant exacerbating factor in this threat landscape. To address these evolving challenges, we introduce CrossPhire: a multimodal deep learning framework with an end-to-end architecture that captures semantic and visual cues from multiple data modalities, while also providing methodological insights for anti-phishing multimodal learning. First, we demonstrate that markup-free semantic text encoding captures linguistic deception patterns more effectively than DOM-based approaches, achieving 96–97% accuracy using textual content alone and providing the strongest single-modality signal through sentence transformers applied to HTML text stripped of structural markup. Second, through controlled comparison of fusion strategies, we show that simple concatenation outperforms a sophisticated gating mechanism so-called Mixture-of-Experts by 0.5–10% when modalities provide complementary, non-redundant security evidence. We validate these insights through rigorous experimentation on five datasets, achieving competitive same-dataset performance (97.96–100%) while demonstrating promising cross-dataset generalization (85–96% accuracy under distribution shift). Additionally, we contribute Phish360, a rigorously curated multimodal benchmark with 10,748 samples addressing quality issues in existing datasets (96.63% unique phishing HTML vs. 16–61% in prior benchmarks), and provide LIME-based explainability tools that decompose predictions into modality-specific contributions. The rapid inference time (0.08 s) and high accuracy results position CrossPhire as a promising solution in the fight against phishing attacks. Full article

To address the common limitations of conventional dual-layer path planning methods, such as slow global convergence, delayed local obstacle avoidance response, and insufficient inter-layer integration, this paper proposes an enhanced collaborative planning framework combining the Improved Dung Beetle Optimizer (IDBO) and the Improved Dynamic Window Approach (IDWA). First, the proposed IDBO solves the problems of population aggregation and unbalanced exploration–exploitation of traditional algorithms by optimizing the initialization strategy and reconstructing the position update mechanism. Second, in the local path planning stage, the IDWA introduces an adaptive evaluation function embedded with obstacle motion prediction and a global path-tracking factor, which breaks through the limitations of traditional local algorithms, such as fixed weights and lack of environmental adaptability, while resolving the contradictions of poor inter-layer coupling and path redundancy in traditional dual-layer frameworks. The results of comparative simulation experiments show that the average path length is reduced by 6.5% and the running time is decreased by 9.1%. This framework effectively overcomes the problems of delayed local response and insufficient inter-layer integration in traditional dual-layer path planning. Full article

While Neural Architecture Search (NAS) has revolutionized the automation of deep learning model design, gradient-based approaches like DARTS often suffer from high computational overheads, the collapse of skip-connections, and optimization instability. To address these limitations, we propose Efficient and Lightweight Differentiable Architecture Search (EL-DARTS). EL-DARTS constructs a compact and redundancy-reduced search space, integrates a partial channel strategy to lower memory usage, employs a Dynamic Coefficient Scheduling Strategy to balance edge importance, and introduces entropy regularization to sharpen operator selection. Experiments on CIFAR-10 and ImageNet demonstrate that EL-DARTS substantially improves both search efficiency and accuracy. Remarkably, it attains a 2.47% error rate on CIFAR-10, requiring merely 0.075 GPU-days for the search. On ImageNet, the discovered architecture achieves a 26.2% top-1 error while strictly adhering to the mobile setting (<600 M MACs). These findings confirm that EL-DARTS effectively stabilizes the search process and pushes the efficiency frontier of differentiable NAS. Full article

This paper presents an architecture and strategy for on-orbit software updating of satellite payload control systems, based on a tightly coupled DSP and FPGA design. The architecture achieves tight coupling between the DSP and FPGA via the XINTF interface, integrating the DSP program and data into the FPGA bitstream. This enables synchronous updating of both chips with a single software package, significantly reducing both uplink data volume and update time. The system features a dual-flash redundant boot design and a mutual supervision mechanism between the DSP and FPGA, enabling cross-monitoring and autonomous reset, thereby significantly enhancing the system’s fault tolerance and reliability in orbit. Experimental results demonstrate a substantial improvement in fault recovery, with the weighted mean recovery time reduced from 27.09 s to 1.56 s, a relative improvement of 94.25% compared to conventional methods. Ground-based environmental tests confirm the system’s stability and engineering viability under extreme space conditions. Full article

Habitat micromodification poses significant challenges to wildlife, necessitating adaptive responses. This study aimed to investigate how such habitat alterations affect the dietary intake and gut microbiota of Père David’s deer (Elaphurus davidianus). A total of 25 fresh fecal samples were collected from Père David’s deer through non-invasive sampling in Tianjin Qilihai Wetland across three distinct phases of habitat micromodification: pre-change (N = 10), under-change (N = 8), and post-change (N = 7). Dietary composition was analyzed via microscopic identification of plant residues, and gut microbiota structure was characterized using 16S rRNA high-throughput sequencing. Results showed that the diet shifted significantly across phases, with 33 plant species from 20 families identified. Meanwhile, the core gut microbiota remained structurally stable. The phyla Firmicutes and Bacteroidota consistently dominated, despite fluctuations in some specific bacterial genera. Functional prediction indicated minimal change in core microbial metabolic pathways. Correlation analysis suggested that key dietary plants were linked to the abundance of specific, functionally relevant microbial taxa. In conclusion, this study demonstrates that the gut microbiota of Père David’s deer exhibits marked resilience to dietary shifts induced by habitat micromodification. This stability is underpinned by functional redundancy within the microbial community and the consistent intake of fibrous plants, representing a key adaptive mechanism. Our findings highlight that integrating non-invasive monitoring of diet and microbiota can effectively assess the adaptive capacity of endangered ungulates to managed habitat change, thereby informing more resilient conservation strategies. Full article