Search Results (364)

Accurate depth image inpainting in complex urban environments remains a critical challenge due to occlusions, reflections, and sensor limitations, which often result in significant data loss. We propose a hybrid deep learning framework that explicitly combines local and global modelling through Convolutional Neural Networks (CNNs) and Transformer modules. The model employs a multi-branch parallel architecture, where the CNN branch captures fine-grained local textures and edges, while the Transformer branch models global semantic structures and long-range dependencies. We introduce an optimized attention mechanism, Agent Attention, which differs from existing efficient/linear attention methods by using learnable proxy tokens tailored for urban scene categories (e.g., façades, sky, ground). A content-guided dynamic fusion module adaptively combines multi-scale features to enhance structural alignment and texture recovery. The frame-work is trained with a composite loss function incorporating pixel accuracy, perceptual similarity, adversarial realism, and structural consistency. Extensive experiments on the Paris StreetView dataset demonstrate that the proposed method achieves state-of-the-art performance, outperforming existing approaches in PSNR, SSIM, and LPIPS metrics. The study highlights the potential of multi-scale modeling for urban depth inpainting and discusses challenges in real-world deployment, ethical considerations, and future directions for multimodal integration. Full article

Dietary polyphenols are recognized as crucial modulators of gastrointestinal motility, holding therapeutic promise for conditions like irritable bowel syndrome, postoperative ileus, and functional dyspepsia. However, their reported effects are heterogeneous, ranging from spasmolytic to prokinetic. This review aims to clarify these inconsistencies by synthesizing experimental evidence on structure–activity relationships and underlying mechanisms. Relevant publications were identified in PubMed and Google Scholar using terms related to polyphenols and gastrointestinal motility. References were selected for relevance, and the narrative review integrates findings from in vitro, ex vivo, in vivo, and clinical studies. Across various experimental models, polyphenols function as multi-target modulators of gastrointestinal smooth muscle. The primary mechanisms identified involve the blockade of voltage-dependent L-type Ca²⁺ channels, activation of K⁺ channels (BK, K_ATP), and modulation of the NO/cGMP and cAMP/PKA pathways. Flavones and multiple flavonols consistently demonstrate spasmolytic activity via Ca²⁺ channel antagonism. In contrast, flavanones engage BK and K_ATP channels to induce membrane hyperpolarization. Complex extracts from plants like ginger and turmeric exhibit mixed pro- or antimotility effects, reflecting the diverse profiles of their constituent compounds. While robust ex vivo pharmacology and some in vivo and human data exist, a high degree of dataset heterogeneity and inconsistent reporting impedes direct translational efforts. Polyphenols are promising multi-mechanistic modulators of gastrointestinal motility with clear structure–activity patterns. To advance their clinical application, future research must focus on establishing standardized in vivo pharmacokinetics, conducting targeted structure–activity studies, employing bioassay-guided fractionation, and designing rigorous clinical trials. Full article

Copper nitride (Cu₃N) is gaining attention as an eco-friendly thin-film semiconductor in a myriad of applications, including storage devices, microelectronic components, photodetectors, and photovoltaic cells. This work presents a detailed optoelectronic study of Cu₃N thin films grown by reactive RF-magnetron sputtering under pure N₂. An overview of the state-of-the-art literature on this material and its potential applications is also provided. The studied films consist of Cu₃N polycrystals with a cubic anti-ReO₃ type structure exhibiting a preferential (100) orientation. Their optical properties across the UV-Vis-NIR spectral range were investigated using a combination of multi-angle spectroscopic ellipsometry, broadband transmission, and reflection measurements. Our model employs a stratified geometrical approach, primarily to capture the depth-dependent compositional variations of the Cu₃N film while also accounting for surface roughness and the underlying glass substrate. The complex dielectric function of the film material is precisely determined through an advanced dispersion model that combines multiple oscillators. By integrating the Tauc–Lorentz, Gaussian, and Drude models, this approach captures the distinct electronic transitions of this polycrystal. This customized optical model allowed us to accurate extract both the indirect (1.83–1.85 eV) and direct (2.38–2.39 eV) bandgaps. Our multifaceted characterization provides one of the most extensive studies of Cu₃N thin films to date, paving the way for optimized device applications and broader utilization of this promising binary semiconductor, and showing its particular potential for photovoltaic given its adequate bandgap energies for solar applications. Full article

In this study, we utilize Gaofen-2 satellite remote sensing images to optimize and enhance the extraction of feature information from rural compounds, addressing key challenges in high-resolution remote sensing analysis: traditional methods struggle to effectively capture long-distance spatial dependencies for scattered rural compounds. To this end, we implement the RS³Mamba+ deep learning model, which introduces the Mamba state space model (SSM) into its auxiliary branching—leveraging Mamba’s sequence modeling advantage to efficiently capture long-range spatial correlations of rural compounds, a critical capability for analyzing sparse rural buildings. This Mamba-assisted branch, combined with multi-directional selective scanning (SS2D) and the enhanced STEM network framework (replacing single 7 × 7 convolution with two-stage 3 × 3 convolutions to reduce information loss), works synergistically with a ResNet-based main branch for local feature extraction. We further introduce a multiscale attention feature fusion mechanism that optimizes feature extraction and fusion, enhances edge contour extraction accuracy in courtyards, and improves the recognition and differentiation of courtyards from regions with complex textures. The feature information of courtyard utilization status is finally extracted using empirical methods. A typical rural area in Weifang City, Shandong Province, is selected as the experimental sample area. Results show that the extraction accuracy reaches an average intersection over union (mIoU) of 79.64% and a Kappa coefficient of 0.7889, improving the F1 score by at least 8.12% and mIoU by 4.83% compared with models such as DeepLabv3+ and Transformer. The algorithm’s efficacy in mitigating false alarms triggered by shadows and intricate textures is particularly salient, underscoring its potential as a potent instrument for the extraction of rural vacancy rates. Full article

To develop truly human-centered automated systems, it is essential to acknowledge that human reasoning is prone to systematic deviations from rational judgment, known as Cognitive Biases. The present study investigated such flawed reasoning in the context of automated driving. In a multi-step study with N = 34 participants, the occurrence of four Cognitive Biases was examined: Truthiness Effect, Automation Bias, Action Bias, and Illusory Control. Additionally, the study explored how the Explainability of the automation’s behavior and the driver’s Mental Model influenced the manifestation of these biases. The findings indicate a notable susceptibility to the Truthiness Effect and Illusory Control, although all biases appeared highly dependent on the specific driving context. Moreover, Explainability strongly impacted the perceived credibility of information and participants’ agreement with the system’s behavior. Given the exploratory nature of the study, this work aims to initiate a discussion on how Cognitive Biases shape human reasoning and decision-making in interactions with automated vehicles. Based on the results, several directions for future research are proposed: (1) investigation of additional cognitive biases, (2) analysis of biases across different levels of automation, (3) exploration of mitigation strategies versus deliberate use of biases, (4) examination of dynamic and context-dependent manifestations, and (5) validation in high-fidelity simulations or real-world settings. Full article

Distribution-level long-term load forecasting with hourly resolution is essential for modern power systems operation, yet it remains challenging due to complex temporal patterns and error accumulation over extended horizons. This study proposes a Multi-Resolution Residual LSTNet framework integrating Discrete Wavelet Transform (DWT), Long Short-Term Memory Networks (LSTNet), and Normalized Linear (NLinear) models for accurate one-year ahead hourly load forecasting. The methodology decomposes load time series into daily, weekly, and monthly components using multi-resolution DWT, applies direct forecasting with LSTNet to capture short-term and long-term dependencies, performs residual correction using NLinear models, and integrates predictions through dynamic weighting mechanisms. Validation using five years of Korean distribution feeder data (2015–2019) demonstrates significant performance improvements over benchmark methods including Autoformer, LSTM, and NLinear, achieving Mean Absolute Error of 0.5771, Mean Absolute Percentage Error of 17.29%, and Huber Loss of 0.2567. The approach effectively mitigates error accumulation common in long-term forecasting while maintaining hourly resolution, providing practical value for demand response, distributed resource control, and infrastructure planning without requiring external variables. Full article

Floating vertical−axis wind turbines present unique advantages for deep−water offshore deployments, but their basin model testing encounters significant challenges in aerodynamic load simulation due to Reynolds scaling effects. While Froude−scaled experiments accurately replicate hydrodynamic behaviors, the drastic reduction in Reynolds numbers at the model scale leads to substantial discrepancies in aerodynamic forces compared to full−scale conditions. This study proposed two methodologies to address these challenges. Fully physical model tests adopt a “physical wind field + rotor model + floating foundation” approach, realistically simulating aerodynamic loads during rotor rotation. Semi−physical model tests employ a “numerical wind field + rotor model + physical floating foundation” configuration, where theoretical aerodynamic loads are obtained through numerical calculations and then reproduced using controllable actuator structures. For fully physical model tests, a blade reconstruction framework integrated airfoil optimization, chord length adjustments, and twist angle modifications through Taylor expansion−based sensitivity analysis. The method achieved thrust coefficient similarity across the operational tip−speed ratio range. For semi−physical tests, a cruciform−arranged rotor system with eight dynamically controlled rotors and constrained thrust allocation algorithms enabled the simultaneous reproduction of periodic streamwise/crosswind thrusts and vertical−axis torque. Numerical case studies demonstrated that the system effectively simulates six−degree−of−freedom aerodynamic loads under turbulent conditions while maintaining thrust variation rates below 9.3% between adjacent time steps. These solutions addressed VAWTs’ distinct aerodynamic complexities, including azimuth−dependent Reynolds number fluctuations and multidirectional force coupling, which conventional methods fail to accommodate. The developed techniques enhanced the fidelity of floating VAWT basin tests, providing critical experimental validation tools for emerging offshore wind technologies. Full article

Accurate natural gas price forecasting is vital for risk management, trading strategies, and policy-making in energy markets. This study proposes and evaluates four hybrid deep learning architectures—CNN-LSTM-Attention, CNN-BiLSTM-Attention, TCN-LSTM-Attention, and TCN-BiLSTM-Attention—integrating convolutional feature extraction, sequential learning, and attention mechanisms. Using Henry Hub and NYMEX datasets, the models are trained on long historical periods and tested under multi-step horizons. The results show that all hybrid models significantly outperform the traditional moving average benchmark, achieving R² values above 95% for one-step-ahead forecasts and maintaining an accuracy of over 87% at longer horizons. CNN-BiLSTM-Attention performs best in short-term prediction due to its ability to capture bidirectional dependencies, while TCN-based models demonstrate greater robustness over extended horizons due to their effective modeling of long-range temporal structures. These findings confirm the advantages of deep learning hybrids in energy forecasting and emphasize the importance of horizon-sensitive evaluation. This study contributes methodological innovation and provides practical insights for market participants, with future directions including the integration of macroeconomic and climatic factors, exploration of advanced architectures such as Transformers, and probabilistic forecasting for uncertainty quantification. Full article

Background: Acute lung injury (ALI) and its severe form, acute respiratory distress syndrome (ARDS), remain major causes of morbidity and mortality, yet no targeted pharmacological therapy is available. Excessive neutrophil and macrophage infiltration drives reactive oxygen species (ROS) production and cytokine release, leading to alveolar–capillary barrier disruption and fatal respiratory failure. Methods: We applied an integrative multi-omics strategy combining single-cell transcriptomics, peripheral blood proteomics, and lung tissue proteomics in a lipopolysaccharide (LPS, 10 mg/kg)-induced mouse ALI model to identify key signaling pathways. Harpagide, an iridoid glycoside identified from our natural compound screen, was evaluated in vivo (40 and 80 mg/kg) and in vitro (0.1–1 mg/mL). Histopathology, oxidative stress markers (SOD, GSH, and MDA), cytokine levels (IL-6 and IL-1β), and signaling proteins (HIF-1α, p-PI3K, p-AKT, Nrf2, and HO-1) were quantitatively assessed. Direct target engagement was probed using surface plasmon resonance (SPR), the cellular thermal shift assay (CETSA), and 100 ns molecular dynamics (MD) simulations. Results: Multi-omics profiling revealed robust activation of HIF-1, PI3K/AKT, and glutathione-metabolism pathways following the LPS challenge, with HIF-1α, VEGFA, and AKT as core regulators. Harpagide treatment significantly reduced lung injury scores by ~45% (p < 0.01), collagen deposition by ~50%, and ROS accumulation by >60% relative to LPS (n = 6). The pro-inflammatory cytokines IL-6 and IL-1β were reduced by 55–70% at the protein level (p < 0.01). Harpagide dose-dependently suppressed HIF-1α and p-AKT expression while enhancing Nrf2 and HO-1 levels (p < 0.05). SPR confirmed direct binding of Harpagide to HIF-1α (KD = 8.73 µM), and the CETSA demonstrated enhanced thermal stability of HIF-1α. MD simulations revealed a stable binding conformation within the inhibitory/C-TAD region after 50 ns. Conclusions: This study reveals convergent immune–microenvironmental regulatory mechanisms across cellular and tissue levels in ALI and demonstrates the protective effects of Harpagide through multi-pathway modulation. These findings offer new insights into the pathogenesis of ALI and support the development of “one-drug, multilayer co-regulation” strategies for systemic inflammatory diseases. Full article

Major stratospheric sudden warmings are key processes in the coupling between the stratosphere and the troposphere, exerting a direct influence on mid-latitude climate variability. This study evaluates projected changes in the frequency of these phenomena during the 2006–2100 period using six high-top general circulation models from the CMIP5 project under the Representative Concentration Pathway scenarios 2.6, 4.5, and 8.5. The analysis combines the full future period with a moving-window approach of 27 and 48 years, compared against both the satellite-era (1979–2005) and extended historical (1958–2005) periods. This methodology reveals that model responses are highly heterogeneous, with alternating periods of significant increases and decreases in event frequency, partially modulated by internal variability. The magnitude and statistical significance of the projected changes strongly depend on the chosen historical reference period, and most models tend to reproduce displacement-type polar vortex events preferentially over split-type events. These results indicate that assessments based solely on multi-model means or long aggregated periods may mask subperiods with robust signals, although some of these may arise by chance given the 5% significance threshold. This underscores the need for temporally resolved analyses to improve the understanding of stratospheric variability and its potential impact on climate predictability. Full article

The accuracy of numerical model outputs strongly depends on the quality of the initial wind field, yet ground observation data are typically sparse and provide incomplete spatial coverage. More importantly, many current mainstream correction models rely on reanalysis grid datasets like ERA5 as the true value, which relies on interpolation calculation, which directly affects the accuracy of the correction results. To address these issues, we propose a new deep learning model, PPWNet. The model directly uses sparse and discretely distributed observation data as the true value, which integrates observation point positions and a physical consistency term to achieve a high-precision corrected wind field. The model design is inspired by biological intelligence. First, observation point positions are encoded as input and observation values are included in the loss function. Second, a parallel dual-branch DenseInception network is employed to extract multi-scale grid features, simulating the hierarchical processing of the biological visual system. Meanwhile, PPWNet references the PointNet architecture and introduces an attention mechanism to efficiently extract features from sparse and irregular observation positions. This mechanism reflects the selective focus of cognitive functions. Furthermore, this paper incorporates physical knowledge into the model optimization process by adding a learned physical consistency term to the loss function, ensuring that the corrected results not only approximate the observations but also adhere to physical laws. Finally, hyperparameters are automatically tuned using the Bayesian TPE algorithm. Experiments demonstrate that PPWNet outperforms both traditional and existing deep learning methods. It reduces the MAE by 38.65% and the RMSE by 28.93%. The corrected wind field shows better agreement with observations in both wind speed and direction, confirming the effectiveness of incorporating position information and a physics-informed approach into deep learning-based wind field correction. Full article

High-resolution pushbroom satellites routinely acquire multi-tenskilometer-scale strips whose vendors’ rational polynomial coefficients (RPCs) exhibit systematic, direction-dependent biases that accumulate downstream when ground control is sparse. This study presents a physically interpretable stripwise extrapolation framework that predicts along- and across-track RPC correlation coefficients for inaccessible segments from an upstream calibration subset. Terrain-independent RPCs were regenerated and residual image-space errors were modeled with weighted least squares using elapsed time, off-nadir evolution, and morphometric descriptors of the target terrain. Gaussian kernel weights favor calibration scenes with a Jarque–Bera-indexed relief similar to the target. When applied to three KOMPSAT-3A panchromatic strips, the approach preserves native scene geometry while transporting calibrated coefficients downstream, reducing positional errors in two strips to <2.8 pixels (~2.0 m at 0.710 m Ground Sample Distance, GSD). The first strip with a stronger attitude drift retains 4.589 pixel along-track errors, indicating the need for wider predictor coverage under aggressive maneuvers. The results clarify the directional error structure with a near-constant across-track bias and low-frequency along-track drift and show that a compact predictor set can stabilize extrapolation without full-block adjustment or dense tie networks. This provides a GCP-efficient alternative to full-block adjustment and enables accurate georeferencing in controlled environments. Full article

With the increasing penetration of wind and photovoltaic (PV) power in modern power systems, accurate power forecasting has become crucial for ensuring grid stability and optimizing dispatch strategies. This study focuses on multiple wind farms and PV plants, where three deep learning models—Long Short-Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM), and a hybrid Transformer–BiLSTM model—are constructed and systematically compared to enhance forecasting accuracy and dynamic responsiveness. First, the predictive performance of each model across different power stations is analyzed. The results reveal that the LSTM model suffers from systematic bias and lag effects in extreme value ranges, while Bi-LSTM demonstrates advantages in mitigating time-lag issues and improving dynamic fitting, achieving on average a 24% improvement in accuracy for wind farms and a 20% improvement for PV plants compared with LSTM. Moreover, the Transformer–BiLSTM model significantly strengthens the ability to capture complex temporal dependencies and extreme power fluctuations. Experimental results indicate that the Transformer–BiLSTM consistently delivers higher forecasting accuracy and stability across all test sites, effectively reducing extreme-value errors and prediction delays. Compared with Bi-LSTM, its average accuracy improves by 19% in wind farms and 35% in PV plants. Finally, this paper discusses the limitations of the current models in terms of multi-source data fusion, outlier handling, and computational efficiency, and outlines directions for future research. The findings provide strong technical support for renewable energy power forecasting, thereby facilitating efficient scheduling and risk management in smart grids. Full article

Cracks are vital warning signs to reflect the structural deterioration in concrete constructions and buildings. However, their diverse and complex morphologies make accurate segmentation challenging. Deep learning-based methods effectively alleviate the low accuracy of traditional methods, while they are limited by the receptive field and computational efficiency, resulting in suboptimal performance. To address this challenging problem, we propose a novel framework termed High-frequency Refined Mamba with Snake Perception Attention module (HFR-Mamba) for more accurate crack segmentation. HFR-Mamba effectively refines Mamba’s global dependency modeling by extracting frequency domain features and the attention mechanism. Specifically, HFR-Mamba consists of the High-frequency Refined Mamba encoder, the Snake Perception Attention (SPA) module, and the Multi-scale Feature Fusion decoder. The encoder uses Discrete Wavelet Transform (DWT) to extract high-frequency texture features and utilizes the Refined Visual State Space (RVSS) module to fuse spatial features and high-frequency components, which effectively refines the global modeling process of Mamba. The SPA module integrates snake convolutions with different directions to filter background noise from the encoder and highlight cracks for the decoder. For the decoder, it adopts a multi-scale feature fusion strategy and a strongly supervised approach to enhance decoding performance. Extensive experiments show HFR-Mamba achieves state-of-the-art performance in IoU, DSC, Recall, Accuracy, and Precision indicators with fewer parameters, validating its effectiveness in crack segmentation. Full article

Predicting the concentrations of air pollutants, particularly PM_2.5, with accuracy and dependability is crucial for protecting human health and preserving a healthy natural environment. This research proposes a deep learning-based, robust prediction system to predict regional PM_2.5 concentrations for the next one to twenty-four hours. To start, the input features of the prediction system are initially screened using a correlation analysis of various air pollutants and meteorological factors. Next, the BiTCN-ISInformer prediction model with a two-branch parallel architecture is constructed. On the one hand, the model improves the probabilistic sparse attention mechanism in the traditional Informer network by optimizing the sampling method from a single sparse sampling to a synergistic mechanism combining sparse sampling and importance sampling, which improves the prediction accuracy and reduces the computational complexity of the model; on the other hand, through the introduction of the bi-directional time-convolutional network (BiTCN) and the design of parallel architecture, the model is able to comprehensively model the short-term fluctuations and long-term trends of the temporal data and effectively increase the inference speed of the model. According to experimental research, the proposed model performs better in terms of prediction accuracy and performance than the most advanced baseline model. In the single-step and multi-step prediction experiments of Shanghai’s PM_2.5 concentration, the proposed model has a root mean square error (RMSE) ranging from 2.010 to 10.029 and a mean absolute error (MAE) ranging from 1.436 to 6.865. As a result, the prediction system proposed in this research shows promise for use in air pollution early warning and prevention. Full article