Search Results (634)

Visible-Infrared Person Re-Identification (VI-ReID) is of crucial importance in applications such as monitoring and security. However, challenges faced from intra-class variations and cross-modal differences are often exacerbated by inaccurate infrared analysis and insufficient structural modeling. To address these issues, we propose Parsing-guided Differential Enhancement Graph Learning (PDEGL), a novel framework that learns discriminative representations through a dual-branch architecture synergizing global feature refinement with part-based structural graph analysis. In particular, we introduce a Differential Infrared Part Enhancement (DIPE) module to correct infrared parsing errors and a Parsing Structural Graph (PSG) module to model high-order topological relationships between body parts for structural consistency matching. Furthermore, we design a Position-sensitive Spatial-Channel Attention (PSCA) module to enhance global feature discriminability. Extensive evaluations on the SYSU-MM01, RegDB, and LLCM datasets demonstrate that our PDEGL method achieves competitive performance. Full article

In the attention-driven environment of online video platforms, understanding the factors that influence content selection and channel subscriptions is crucial for creators, marketers, and platform managers. This study investigates how thumbnails, view counts, video length, genre, and the number of advertisements affect user decision-making on YouTube. Grounded in random utility theory and reference-dependent preference theory, this study conducted a choice experiment with 525 respondents and employed a combined model of rank-ordered and binary logit methods to analyze viewing and subscription behaviors. The results indicate a significant preference for thumbnails with subtitles and shorter videos. Notably, we found evidence of reference-dependent effects, whereby a higher-than-expected number of ads decreased viewing probability, while a lower-than-expected number significantly increased subscription probability. This study advances our understanding of the factors that influence user behavior on social media, specifically in terms of viewing and subscribing, and empirically supports prospect theory in the online advertising market. Our findings offer both theoretical and practical insights into optimizing video content and monetization strategies in competitive social media markets. Full article

Rice blast caused by Magnaporthe oryzae is a major cause of yield reductions and quality deterioration in rice. Therefore, early detection of the disease is necessary for controlling the spread of rice blast. This study proposed a dual-channel feature fusion model (DCFM) to achieve effective identification of rice blast. The DCFM model extracted spectral features using successive projection algorithm (SPA), random frog (RFrog), and competitive adaptive reweighted sampling (CARS), and extracted spatial features from spectral images using MobileNetV2 combined with the convolutional block attention module (CBAM). Then, these features were fused using the feature fusion adaptive conditioning module in DCFM and input into the fully connected layer for disease identification. The results show that the model combining spectral and spatial features was superior to the classification models based on single features for rice blast detection, with OA and Kappa higher than 90% and 88%, respectively. The DCFM model based on SPA screening obtained the best results, with an OA of 96.72% and a Kappa of 95.97%. Overall, this study enables the early and accurate identification of rice blast, providing a rapid and reliable method for rice disease monitoring and management. It also offers a valuable reference for the detection of other crop diseases. Full article

As Android malware grows increasingly sophisticated, traditional detection methods struggle to keep pace, creating an urgent need for robust, interpretable, and real-time solutions to safeguard mobile ecosystems. This study introduces YoloMal-XAI, a novel deep learning framework that transforms Android application files into RGB image representations by mapping DEX (Dalvik Executable), Manifest.xml, and Resources.arsc files to distinct color channels. Evaluated on the CICMalDroid2020 dataset using YOLO11 pretrained classification models, YoloMal-XAI achieves 99.87% accuracy in binary classification and 99.56% in multi-class classification (Adware, Banking, Riskware, SMS, and Benign). Compared to ResNet-50, GoogLeNet, and MobileNetV2, YOLO11 offers competitive accuracy with at least 7× faster training over 100 epochs. Against YOLOv8, YOLO11 achieves comparable or superior accuracy while reducing training time by up to 3.5×. Cross-corpus validation using Drebin and CICAndMal2017 further confirms the model’s generalization capability on previously unseen malware. An ablation study highlights the value of integrating DEX, Manifest, and Resources components, with the full RGB configuration consistently delivering the best performance. Explainable AI (XAI) techniques—Grad-CAM, Grad-CAM++, Eigen-CAM, and HiRes-CAM—are employed to interpret model decisions, revealing the DEX segment as the most influential component. These results establish YoloMal-XAI as a scalable, efficient, and interpretable framework for Android malware detection, with strong potential for future deployment on resource-constrained mobile devices. Full article

This research investigates enhancing the performance of an Archimedes screw-type hydrokinetic turbine (ASHT). A 3D transient computational model employing the six degrees of freedom (6-DOF) methodology within the ANSYS Fluent software 2022 R1, was selected for this purpose. A central composite design (CCD) methodology was applied within the response surface methodology (RSM) to enhance the turbine’s power coefficient ($C_p$). Key independent factors, including blade length (L), blade inclination angle ($\gamma$), and external diameter ($D_e$), were systematically varied to determine their optimal values. The optimization process yielded a maximum $C_p$ of 0.337 for L, $\gamma$, and $D_e$ values of 168.921 mm, 51.341°, and 245.645 mm, respectively. Experimental validation was conducted in a hydraulic channel, yielding results that demonstrated a strong correlation with the numerical predictions. This research underscores the importance of geometric design optimization in improving the energy capture efficiency of the ASHT, contributing to its potential viability as a competitive renewable energy solution in the pre-commercial phase of development. Full article

With an economic model characterized by high energy consumption and low efficiency, China is facing serious energy shortages and environmental problems. However, education, as the cornerstone of social progress, has been overlooked in its role in improving energy efficiency. This study aims to enhance our understanding of the impact of educational competitiveness on urban green total factor energy efficiency (GTFEE), helping policymakers to achieve sustainable urban development. This study utilizes panel data from 20 major Chinese cities spanning from 2012 to 2022 and applies a two-way fixed effects model to investigate the relationship and pathways of educational competitiveness (Ec) on GTFEE. Our results show that the Ec index can enhance the major urban GTFEE. Among them, educational resource competitiveness, input competitiveness, efficiency competitiveness, and sustainable competitiveness can all enhance urban GTFEE, but the coefficient of the educational scale is not significant. In addition, Ec can effectively improve GTFEE by promoting green technological innovation, alleviating human resource mismatch, and driving industrial structure upgrading. Furthermore, the impact of Ec on GTFEE shows significant regional heterogeneity, with its effect weakening from the eastern coastal areas to the western inland regions. Full article

Tunneling modes in all-dielectric one-dimensional photonic crystals can be utilized for multi-channel filtering. However, these tunneling modes generally blue shift upon increasing the incident angle. When hyperbolic metamaterials are introduced into one-dimensional photonic crystals, the competition between the propagation phase shifts in the dielectric materials and hyperbolic metamaterials can result in different angle dependencies, including blue shift, abnormal zero shift, and abnormal red shift. When the reduction in the propagation phase in the dielectric layer exceeds the increment in the propagation phase in the hyperbolic metamaterial, the tunneling modes are blue-shifted; conversely, when the phase increment in the hyperbolic metamaterial exceeds the phase reduction in the dielectric layer, the tunneling modes are abnormally red-shifted. When the phase changes in the two materials are the same, the tunneling modes are angle independent. In this study, we investigated the multiple filtering effects of one-dimensional photonic structures composed of hyperbolic metamaterials. These composed structures exhibited multiple tunneling modes based on one-, two-, or three-angle dependencies and can be applied in novel optical devices with different angle-dependence requirements. Full article

Underwater inspection images of hydraulic structures often suffer from haze, severe color distortion, low contrast, and blurred textures, impairing the accuracy of automated crack, spalling, and corrosion detection. However, many existing enhancement methods fail to preserve structural details and suppress noise in turbid environments. To address these limitations, we propose a compact image enhancement framework called Wavelet Fusion with Sobel-based Weighting (WWSF). This method first corrects global color and luminance distributions using multiscale Retinex and gamma mapping, followed by local contrast enhancement via CLAHE in the L channel of the CIELAB color space. Two preliminarily corrected images are decomposed using discrete wavelet transform (DWT); low-frequency bands are fused based on maximum energy, while high-frequency bands are adaptively weighted by Sobel edge energy to highlight structural features and suppress background noise. The enhanced image is reconstructed via inverse DWT. Experiments on real-world sluice gate datasets demonstrate that WWSF outperforms six state-of-the-art methods, achieving the highest scores on UIQM and AG while remaining competitive on entropy (EN). Moreover, the method retains strong robustness under high turbidity conditions (T ≥ 35 NTU), producing sharper edges, more faithful color representation, and improved texture clarity. These results indicate that WWSF is an effective preprocessing tool for downstream tasks such as segmentation, defect classification, and condition assessment of hydraulic infrastructure in complex underwater environments. Full article

This study systematically investigates the effects of trap concentration on self-heating and terahertz (THz) responses in GaN HEMTs using Sentaurus TCAD. Traps, inherently unavoidable in semiconductors, can be strategically introduced to engineer specific energy levels that establish competitive dynamics between the electron momentum relaxation time and the carrier lifetime. A simulation-based exploration of this mechanism provides significant scientific value for enhancing device performance through self-heating mitigation and THz response optimization. An AlGaN/GaN heterojunction HEMT model was established, with trap concentrations ranging from 0 to $5\times {10}^{17} {\mathrm{c}\mathrm{m}}^{-3}$. The analysis reveals that traps significantly enhance channel current (achieving 3× gain at $1\times 10^{17} {\mathrm{c}\mathrm{m}}^{-3}$) via new energy levels that prolong carrier lifetime. However, elevated trap concentrations (>$1\times {10}^{16} {\mathrm{c}\mathrm{m}}^{-3}$) exacerbate self-heating-induced current collapse, reducing the min-to-max current ratio to 0.9158. In THz response characterization, devices exhibit a distinct DC component (${\mathrm{U}}_{\mathrm{d}\mathrm{c}}$) under non-resonant detection ($\mathsf{\omega }\mathsf{\tau }\ll 1$). At a trap concentration of $1\times {10}^{15} {\mathrm{c}\mathrm{m}}^{-3}$, ${\mathrm{U}}_{\mathrm{d}\mathrm{c}}$ peaks at $0.12 \mathrm{V}$ when ${\mathrm{V}}_{\mathrm{g}}\mathrm{D}\mathrm{C}=-7.8 \mathrm{V}$. Compared to trap-free devices, a maximum response attenuation of 64.89% occurs at ${\mathrm{V}}_{\mathrm{g}}\mathrm{D}\mathrm{C}=-4.9 \mathrm{V}$. Furthermore, ${\mathrm{U}}_{\mathrm{d}\mathrm{c}}$ demonstrates non-monotonic behavior with concentration, showing local maxima at $4\times {10}^{15} {\mathrm{c}\mathrm{m}}^{-3}$ and $7\times {10}^{15} {\mathrm{c}\mathrm{m}}^{-3}$, attributed to plasma wave damping and temperature-gradient-induced electric field variations. This research establishes trap engineering guidelines for GaN HEMTs: a concentration of $4\times {10}^{15} {\mathrm{c}\mathrm{m}}^{-3}$ optimally enhances conductivity while minimizing adverse impacts on both self-heating and the THz response, making it particularly suitable for high-sensitivity terahertz detectors. Full article

Target detection is the key technology of the apple-picking robot. To overcome the limitations of existing apple target detection methods, including low recognition accuracy of multi-species apples in complex orchard environments and a complex network architecture that occupies large memory, a lightweight apple recognition model based on the improved YOLO11 model was proposed, named YOLO-WAS model. The model aims to achieve efficient and accurate automatic multi-species apple identification while reducing computational resource consumption and facilitating real-time applications on low-power devices. First, the study constructed a high-quality multi-species apple dataset and improved the complexity and diversity of the dataset through various data enhancement techniques. The YOLO-WAS model replaced the ordinary convolution module of YOLO11 with the Adown module proposed in YOLOv9, the backbone C3K2 module combined with Wavelet Transform Convolution (WTConv), and the spatial and channel synergistic attention module Self-Calibrated Spatial Attention (SCSA) combined with the C2PSA attention mechanism to form the C2PSA_SCSA module was also introduced. Through these improvements, the model not only ensured lightweight but also significantly improved performance. Experimental results show that the proposed YOLO-WAS model achieves a precision (P) of 0.958, a recall (R) of 0.921, and mean average precision at IoU threshold of 0.5 (mAP@50) of 0.970 and mean average precision from IoU threshold of 0.5 to 0.95 with step 0.05 (mAP@50:95) of 0.835. Compared to the baseline model, the YOLO-WAS exhibits reduced computational complexity, with the number of parameters and floating-point operations decreased by 22.8% and 20.6%, respectively. These results demonstrate that the model performs competitively in apple detection tasks and holds potential to meet real-time detection requirements in resource-constrained environments, thereby contributing to the advancement of automated orchard management. Full article

Motor Imagery (MI) based Brain Computer Interfaces (BCIs) have promising applications in neurorehabilitation for individuals who have lost mobility and control over parts of their body due to brain injuries, such as stroke patients. Accurately classifying MI tasks is essential for effective BCI performance, but this task remains challenging due to the complex and non-stationary nature of EEG signals. This study aims to improve the classification of left and right-hand MI tasks by utilizing high-resolution time-frequency features extracted from EEG signals, enhanced with deep learning-based data augmentation techniques. We propose a novel deep learning framework named the Generalized Wavelet Transform-based Deep Convolutional Network (GDC-Net), which integrates multiple components. First, EEG signals recorded from the C3, C4, and Cz channels are transformed into detailed time-frequency representations using the Generalized Morse Wavelet Transform (GMWT). The selected features are then expanded using a Deep Convolutional Generative Adversarial Network (DCGAN) to generate additional synthetic data and address data scarcity. Finally, the augmented feature maps data are subsequently fed into a hybrid CNN-LSTM architecture, enabling both spatial and temporal feature learning for improved classification. The proposed approach is evaluated on the BCI Competition IV dataset 2b. Experimental results showed that the mean classification accuracy and Kappa value are 89.24% and 0.784, respectively, making them the highest compared to the state-of-the-art algorithms. The integration of GMWT and DCGAN significantly enhances feature quality and model generalization, thereby improving classification performance. These findings demonstrate that GDC-Net delivers superior MI classification performance by effectively capturing high-resolution time-frequency dynamics and enhancing data diversity. This approach holds strong potential for advancing MI-based BCI applications, especially in assistive and rehabilitation technologies. Full article

As communication technologies continue to evolve, Reconfigurable Intelligent Surfaces (RISs) have become a crucial and highly potential technology for sixth-generation (6G) mobile communication systems. Their key competitive advantages lie in their cost-effectiveness, minimal power consumption, and simple deployment. To address the limitations of current communication paradigms, this study innovatively integrates RIS technology into vehicle-to-vehicle (V2V) communication systems. Current methodologies fail to comprehensively elucidate the transmission principles underlying RIS-assisted V2V fluid antenna system (FAS) communications. The current channel characteristic analysis techniques and modeling theories struggle to achieve a balance between computational accuracy and computational complexity. To overcome these problems, this study systematically constructed a multipath sub-channel model in RIS-assisted V2V communication. Combining detailed simulation with theoretical analysis, a reliable parametric channel statistical model was established. This progress successfully overcame the main obstacle of the traditional RIS channel modeling method, which was unable to coordinate accuracy and efficiency. Full article

Climate change risk has significant impacts on corporate financial activities. Using firm-level data from A-share listed companies in China from 2010 to 2022, we examine how climate risk affects corporate debt financing capacity. We find that climate change risk significantly weakens firms’ ability to raise debt, leading to lower leverage and higher financing costs. These results remain robust across various checks for endogeneity and alternative specifications. We also show that reducing corporate carbon emission intensity can mitigate the negative impact of climate risk on debt financing, suggesting that supply-side credit policies are more effective than demand-side capital structure choices. Furthermore, we identify three channels through which climate risk impairs debt capacity: reduced competitiveness, increased default risk, and diminished resilience. Our heterogeneity analysis reveals that these adverse effects are more pronounced for non-state-owned firms, firms with weaker internal controls, and companies in highly financialized regions, and during periods of heightened environmental uncertainty. We also apply textual analysis and machine learning to the measurement of climate change risks, partially mitigating the geographic biases and single-dimensional shortcomings inherent in macro-level indicators, thus enriching the quantitative research on climate change risks. These findings provide valuable insights for policymakers and financial institutions in promoting corporate green transition, guiding capital allocation, and supporting sustainable development. Full article

Green and low-carbon development of supply chains represents a practical approach to addressing climate change and enhancing corporate competitiveness. From the perspective of the relationship between policy subsidies and channel power structures, this paper constructs Stackelberg game models under four different scenarios to conduct theoretical analyses of the optimal strategies, supported by numerical simulations. The research findings reveal the following. (1) Under the product subsidy policy, the enhancement of consumers’ green preference will lead to a green premium, and in the case of the technology subsidy policy, consumers’ green preference will inhibit wholesale prices and retail prices. However, there is a threshold in the manufacturer-led case, and a “green premium” is also claimed when this threshold is exceeded. (2) The effects of the product subsidy policy and the green technology level subsidy policy on prices are opposite, where an increase in the product subsidy will increase the wholesale price and retail price, while an increase in the green technology level subsidy will reduce the wholesale price. The technology subsidy policy has a more significant effect on the promotion of green technology. (3) The power of supply chain channels will directly affect corporate profits, and the leader of the supply chain often has higher profits. Compared with product subsidies, technology subsidies can inhibit the channel power of retailers. Full article

In recent years, remote sensing imagery has become indispensable for applications such as environmental monitoring, land use classification, and urban planning. However, the physical constraints of satellite imaging systems frequently limit the spatial resolution of these images, impeding the extraction of fine-grained information critical to downstream tasks. Super-resolution (SR) techniques thus emerge as a pivotal solution to enhance the spatial fidelity of remote sensing images via computational approaches. While deep learning-based SR methods have advanced reconstruction accuracy, their high computational complexity and large parameter counts restrict practical deployment in real-world remote sensing scenarios—particularly on edge or low-power devices. To address this gap, we propose LSANet, a lightweight SR network customized for remote sensing imagery. The core of LSANet is the large separable kernel attention mechanism, which efficiently expands the receptive field while retaining low computational overhead. By integrating this mechanism into an enhanced residual feature distillation module, the network captures long-range dependencies more effectively than traditional shallow residual blocks. Additionally, a residual feature enhancement module, leveraging contrast-aware channel attention and hierarchical skip connections, strengthens the extraction and integration of multi-level discriminative features. This design preserves fine textures and ensures smooth information propagation across the network. Extensive experiments on public datasets such as UC Merced Land Use and NWPU-RESISC45 demonstrate LSANet’s competitive or superior performance compared to state-of-the-art methods. On the UC Merced Land Use dataset, LSANet achieves a PSNR of 34.33, outperforming the best-baseline HSENet with its PSNR of 34.23 by 0.1. For SSIM, LSANet reaches 0.9328, closely matching HSENet’s 0.9332 while demonstrating excellent metric-balancing performance. On the NWPU-RESISC45 dataset, LSANet attains a PSNR of 35.02, marking a significant improvement over prior methods, and an SSIM of 0.9305, maintaining strong competitiveness. These results, combined with the notable reduction in parameters and floating-point operations, highlight the superiority of LSANet in remote sensing image super-resolution tasks. Full article