Search Results (560)

Underwater sonar object detection is challenging because targets are often small, boundaries are blurred, background clutter is strong, and labeled sonar data are limited. To address these issues, we propose T2C-DETR, a detector built on RT-DETR with three task-oriented improvements: (i) a Transformer–Convolution dual-channel backbone (TCDCNet) for complementary global-context and local-detail modeling, (ii) a Noise Filtering Module (NFM) inserted before neck fusion to suppress noise-dominated activations, and (iii) a stage-wise transfer-learning strategy tailored to small sonar datasets. We evaluate the method under three pre-training sources (COCO 2017, DOTA, and an infrared dataset) and then fine-tune on a self-built sonar dataset. Experimental results show that T2C-DETR achieves AP50 of 97.8%, 98.2%, and 98.5% at 72–73 FPS, consistently outperforming the RT-DETR baseline, YOLOv5-Imp, and MLFFNet in the accuracy–speed trade-off. These results indicate that combining global–local representation learning with targeted noise suppression is effective for practical real-time sonar detection. Full article

This study evaluates the application of deep learning techniques to the design of a microwave transmitter–receiver system operating in the 5G mid-band. The proposed architecture consists of four stages—signal generation, amplification, mixing, and filtering—each initially designed using conventional microwave methods and subsequently integrated into a complete transceiver. Simulation data were generated and component-specific convolutional neural networks (CNNs) were implemented in Python using TensorFlow/Keras. Across all models, an average error reduction exceeding 90% was achieved, with most networks converging after the third training cycle. System-level integration shows that the baseline design achieved a transmitted power of −32.637 dBm and a gain of 1.116 dB, while the deep learning-based design yielded −33.912 dBm and 0.738 dB. Additional analysis of S-parameters confirms acceptable impedance matching and a frequency response of around 3.5 GHz. These results illustrate that deep learning provides an effective complementary methodology for multi-component microwave system modeling and optimization in 5G applications. Full article

Utility-scale photovoltaic (PV) plants rely on supervisory control and data acquisition (SCADA) streams for performance verification, yet high-frequency measurements are routinely affected by sensor dropouts, intermittency, and operating-state transitions that bias regression-based capacity estimates. This study evaluates a reproducible SCADA processing workflow for capacity-style reporting and a complementary soiling–clean temperature prediction model using data from a documented October 2022 test window (5 s SCADA aggregated to 1 min). The following three filtering approaches are compared: (i) naïve thresholds (Baseline A), (ii) deterministic stability screening using ramp-rate and rolling-variability constraints (Baseline B), and (iii) an optional residual-based outlier trimming step (Method C). Capacity is estimated via a multivariate regression evaluated on a fixed-size reporting-condition subset (RC197) with day-coverage constraints. All methods achieved high fit quality on RC197 ($R^2\approx 0.99$), with Baseline B improving error and uncertainty over Baseline A (RMSE 2.05 vs. 2.18 MW; $U_{95}$ 0.97% vs. 1.03%) while preserving day coverage; Method C yielded the lowest in-sample RMSE (1.89 MW) but reduced day coverage. For temperature prediction, a baseline-plus-residual learning formulation substantially improved leave-one-day-out performance, reducing MAE/RMSE from 2.99/3.76 °C to 1.43/1.80 °C and increasing $R^2$ from 0.60 to 0.91. The results highlight trade-offs between fit tightness and representativeness in capacity-style filtering and demonstrate residual learning is an effective approach for SCADA-based thermal characterization. Full article

4D millimeter-wave radar provides high-precision ranging capability and exhibits strong robustness under adverse weather and low-visibility conditions, but its point clouds are relatively sparse and suffer from severe elevation-angle measurement noise. Monocular cameras, by contrast, provide rich semantic information and high recall, yet are fundamentally limited by scale ambiguity. To exploit the complementary characteristics of these two sensors, this paper proposes a radar-camera fusion 3D multi-object tracking framework that does not rely on complex 3D annotated data. First, on the radar signal-processing side, a Gaussian distribution-based adaptive angle compression method and IMU-based velocity compensation are introduced to effectively suppress measurement noise, and an improved DBSCAN clustering scheme with recursive cluster splitting and historical static-box guidance is employed to generate high-quality radar detections. Second, a disparity-domain metric depth recovery method is proposed. This method uses filtered radar points as sparse metric anchors, performs robust fitting with RANSAC, and applies Kalman filtering for temporal smoothing, thereby converting the relative depth output of the visual foundation model Depth Anything V2 into metric depth. Finally, a hierarchical fusion strategy is designed at both the detection and tracking levels to achieve stable cross-modal state association. Experimental results on a self-collected dataset show that the proposed method achieves an overall MOTA of 77.93%, outperforming single-modality baselines and other comparison methods by 11 to 31 percentage points. This study provides an effective solution for low-cost and robust environment perception in complex dynamic scenarios. Full article

Cascade dams fragment river habitats, but how seasonal hydrology modulates the biogeography and assembly of fish gut microbiota remains unclear. We surveyed gut bacterial communities of the omnivorous fish Saurogobio punctatus across 10 reaches separated by cascade dams in the Qijiang River during the wet (summer) and dry (winter) seasons using 16S rRNA gene amplicon sequencing. Sampling was synchronized among reaches to minimize temporal variability. Winter exhibited stronger differentiation among reaches and a steeper distance–decay pattern, and reach-scale environmental heterogeneity (especially dissolved inorganic nitrogen) was more stable under weak hydrodynamics. Null model analyses showed that stochastic processes dominated in summer, with dispersal-related processes and drift being prominent under high connectivity, whereas deterministic assembly increased in winter and was mainly associated with homogeneous selection. Compositionality-aware differential abundance analysis (ANCOM-BC2) identified 409 genera with a significant seasonal differential abundance after adjusting for reach (FDR q < 0.05). Random forest classification, used as a complementary prediction-oriented feature-ranking analysis, indicated higher reach discriminability in winter, with Nitrospirota ranking among the top features. PLS-PM indicated that α-diversity had the strongest direct association with β-diversity in the specified model, whereas spatial and environmental effects were linked to β-diversity mainly through indirect, α-diversity-mediated pathways. Biologically, α-diversity may reflect an integrative summary of the within-gut taxon pool shaped by host filtering and environmentally derived inputs (e.g., diet- and habitat-associated sources), which can influence the magnitude of between-reach compositional turnover. Together, these results show that seasonal hydrological regimes tune spatial turnover and assembly of fish gut microbiota in cascade-regulated rivers. Full article

This work presents a portable polarimetric multispectral imaging (PMSI) system operating in the visible to shortwave infrared range (VNIR–SWIR: 400–1700 nm) and its application to target detection, discrimination from aquatic backgrounds, and polymer identification. The instrument integrates two synchronized cameras with motorized bandpass filters and piezoelectric polarization control, enabling the acquisition of 48 wavelength–polarization measurements per capture. This configuration allows the extraction of both intensity-based and polarimetric features, including the degree of linear polarization (DoLP). A complete radiometric and polarimetric calibration framework is implemented, encompassing system response characterization, polarization-dependent gain correction, and reflectance normalization under variable illumination. Experiments conducted on a representative set of 16 polymer materials show that polarimetric information consistently improves class separability compared to intensity-only features, with a mean gain of 6.9 (95% CI: 6.35–8.47). Although the correlation between intensity- and DoLP-based separability is moderate (r = 0.44), the results indicate complementary identification capability. Material recoverability was further evaluated using spectral unmixing techniques (VCA, N-FINDR, and PPI), with VCA offering the best accuracy–complexity trade-off on the calibrated Stokes reflectance dataset. Despite these gains, identification among chemically similar polyethylene variants remains challenging due to limited spectral and polarimetric contrast. An underwater detectability study under natural illumination reveals strong wavelength-dependent constraints: SWIR penetration is limited to 4 cm, whereas VNIR bands (430–550 nm) preserve detectability up to 20 cm, with DoLP enhancing edge visibility. These results motivate future validation in more complex aquatic conditions and with increased spectral dimensionality. Full article

Hyperspectral image (HSI) classification has achieved substantial progress with deep learning. However, existing methods still underexploit frequency-domain information, particularly the complementary roles of high- and low-frequency components. The recently proposed Kolmogorov-Arnold Network (KAN) shows strong nonlinear feature extraction ability for HSI classification, but its lack of frequency-domain learning and reliance on B-spline activation functions often causes unstable training and convergence issues. To address these limitations, this study introduces a Frequency-aware Filtering Activation-based KAN (FFAKAN) for HSI classification. In this framework, the conventional B-spline activation functions in KAN are replaced with learnable high-pass and low-pass spatial filters, enabling explicit frequency decomposition while preserving spectral sequence modeling capacity. Specifically, the proposed framework includes three modules: spectral-spatial feature embedding (S2FE), adaptive filtering KAN (AFKAN), and sequence feature extraction (SeqFE) modules. First, the S2FE module generates highly discriminative feature representations, providing a strong foundation for subsequent processing. Second, the AFKAN module, serving as the core component, employs learnable cutoff frequencies together with cosine-based smooth transition functions to achieve physically interpretable high-low frequency separation, adaptively capturing fine-grained details and structural characteristics in HSI data. Finally, the SeqFE module leverages multi-layer stacked 3D convolutions to perform deep spectral-spatial correlation modeling, extracting high-level discriminative joint features for the classification task. Experiments on four public HSI datasets demonstrate that FFAKAN consistently outperforms state-of-the-art methods. Overall, the proposed method achieves significant performance gains, with maximum improvements of 6.82%, 1.83%, 4.35%, and 5.93% compared with conventional approaches. Moreover, compared with strong baseline models, FFAKAN further improves the overall accuracy by 1.70%, 0.10%, 0.02%, and 0.37%, respectively. These results clearly demonstrate the effectiveness, robustness, and superior generalization capability of the proposed method across different datasets. This study introduces a new paradigm that incorporates physically interpretable frequency-domain priors, showing strong adaptability and promising potential in complex land-cover scenarios. Full article

High-rise elevator group control systems operate under pronounced nonstationarity during commuting peaks, post-event surges, and capacity degradation, where the waiting time distribution becomes right-tail heavy and stresses service-level agreements (SLAs) defined by coverage and high-quantile targets. At the same time, the time-of-use tariffs and carbon constraints sharpen the tension between peak-power control, energy savings, and service capacity. This paper proposes a two-layer resilience scheduling framework that integrates queueing-based planning with safe reinforcement learning (RL) fine-tuning. In the planning layer, parsimonious queueing approximations and scenario-based evaluation construct a finite set of implementable mode cards and emergency switching cards; Sample Average Approximation (SAA) combined with Conditional Value-at-Risk (CVaR) constraints filter candidates to enforce tail-risk-aware service limits while keeping power demand within a prescribed envelope. In the execution layer, online dispatch is formulated as a constrained Markov decision process; within the planning layer limits, action masking and Lagrangian safe RL learn small adaptive adjustments to suppress tail-waiting risk and improve recovery dynamics without increasing peak-power commitments. The experiments under morning peaks and post-event surges confirm tail risk reduction and accelerated recovery. For partial outages, the framework prioritizes SLA coverage and recovery speed, accepting a bounded increase in tail risk as a manageable trade-off. Throughout all tests, peak power remains within the prescribed limits. Improvements persist across random seeds and demand fluctuations, indicating distributional robustness and cross-scenario generalization. Ablation studies further reveal complementary roles: removing the planning layer CVaR screening worsens tail performance, while removing the execution layer action masking increases constraint violations and destabilizes recovery. Full article

Background: Tumor neoantigens are key targets for personalized vaccines and T-cell therapies, yet most pipelines focus on neoantigens derived from SNV/small indel and often yield a limited number of high-quality candidates. SVs are prevalent in tumors and can generate novel chimeric sequences and neopeptides, making them a promising additional source of neoantigens. However, SV-derived neoantigen prediction remains challenging due to breakpoint uncertainty, isoform-dependent coding inference, and limited integration of multi-dimensional evidence and reproducibility. Methods: We developed SVNeoPP (Structural Variant Neoantigen Prediction and Prioritization), an end-to-end workflow for SV-derived neoantigen analysis. SVNeoPP takes WGS and RNA-seq as inputs, performs SV calling and annotation, and reconstructs altered transcripts and coding sequences in a traceable, isoform-aware manner to generate candidate peptides. Candidates are prescreened by integrating antigen-processing features with HLA binding prediction, and then hierarchically filtered and prioritized based on transcript expression, LC–MS/MS proteomics evidence, immunogenicity predictions, and sequence similarity to experimentally validated neoantigen databases. SVNeoPP is implemented in Snakemake to enable modular extension, checkpoint-based restarts, and end-to-end reproducibility. Results: Using a hepatocellular carcinoma (HCC) multi-omics dataset as a proof of concept, we demonstrated the performance of SVNeoPP and obtained a high-priority shortlist of candidate peptides. Compared with other methods, SVNeoPP substantially expanded the candidate search space for SV-derived neoantigens and showed more favorable distributions of antigen-processing and HLA binding features. Conclusions: SVNeoPP provides a reusable, traceable, and interpretable multi-dimensional evidence-driven framework for SV-derived neoantigens. As a complementary module to SNV/small-indel pipelines, it broadens the neoantigen candidate repertoire and generates ranked candidates with interpretable evidence to facilitate downstream prioritization and decision-making. Full article

Conventional image stitching pipelines predominantly rely on homographic alignment, whose planar assumption often breaks down in dual-camera configurations capturing non-planar scenes, producing geometric warping, bulging, and structural distortion. To address these limitations, this paper presents SENA (Seamlessly Natural), a geometry-driven image stitching approach with three complementary contributions. First, we propose a hierarchical affine-based warping strategy that combines global affine initialization, local affine refinement, and a smooth free-form deformation field regulated by seamguard adaptive smoothing. This multi-scale design preserves local shape, parallelism, and aspect ratios, thereby reducing the hallucinated distortions commonly associated with homography-based models. Second, SENA incorporates a geometry-driven adequate zone detection mechanism that identifies regions with reduced parallax directly from the disparity consistency of correspondences filtered by RANSAC, without relying on semantic segmentation or depth estimation. Third, within this zone, anchor-based seamline cutting and segmentation enforce one-to-one geometric correspondence between image pairs, reducing ghosting and smearing artifacts. Extensive experiments demonstrate that SENA achieves 26.2 dB PSNR and 0.84 SSIM, obtains the lowest BRISQUE score (33.4) among compared methods, and reduces runtime by 79% on average across resolutions. These results confirm improved structural fidelity and computational efficiency while maintaining competitive alignment accuracy. Full article

Space-based multi-target passive tracking is critical for space situational awareness, but faces severe challenges due to the limited field-of-view (FoV) and directional ambiguity of onboard sensors. These constraints often lead to target loss, poor observability, and decreased estimation accuracy. To address these issues, different fusion architectures have been explored. While centralized measurement-level fusion offers superior accuracy for estimating target states, distributed estimation-level fusion provides greater reliability for estimating the number of targets. To adaptively leverage these two complementary strengths, a dynamic hierarchical fusion method through real-time optimization of the fusion topology is proposed. Specifically, at each decision epoch, sensor nodes are dynamically partitioned into local fusion nodes (LFNs) and detection-only nodes (DONs). Each LFN receives measurements from selected DONs and executes an iterated-correction Gaussian-mixture probability hypothesis density filter. Subsequently, LFNs share and fuse their estimates using the intensity-dependent arithmetic average fusion. This dynamic process is achieved by applying a sensor management scheme based on partially observable Markov decision process (POMDP). To ensure accurate cardinality estimation, the reward function in POMDP utilizes the posterior expected number of targets. The resultant optimization is efficiently solved using a binary particle swarm optimization algorithm. Numerical and hardware-in-the-loop simulations demonstrate the effectiveness of the proposed method in balancing the accuracy of target number and state estimation. Full article

Image restoration has wide applications in the field of computer vision, yet existing methods suffer from limitations. CNNs struggle to capture long-range dependencies, while transformers exhibit insufficient performance in handling local details and high computational complexity. Additionally, existing dual-branch networks fail to define a clear dominant–auxiliary role between branches, leading to redundancy and high computational costs. This paper proposes a Global Modulated Unbalanced Dual-Branch Network (GMUD-Net), which innovatively adopts an unbalanced structure with a CNN as the main branch and a transformer as the auxiliary branch. Specifically, the CNN branch achieves strong restoration capability by integrating the global–local hybrid backbone block (GLBB) and the frequency-based global attention module (FGAM). As the key building block in the CNN branch, GLBB integrates a local backbone branch, a global Fourier branch, and a residual branch to fuse local details with global context. Meanwhile, FGAM leverages the fast Fourier transform at the bottleneck to enhance cross-channel interaction and improve global restoration performance. In addition, the lightweight transformer branch employs efficient cross-channel attention to provide complementary global cues, which are filtered and injected into the CNN branch via the global attention guidance block (GAG). These designs integrate the advantages of both CNNs and transformers while significantly reducing computational burden, offering a new paradigm to address the limitations of traditional dual-branch architectures. Experimental results demonstrate that compared with existing algorithms, the proposed method achieves state-of-the-art or highly competitive performance in both quantitative evaluations and qualitative results across nine datasets. Full article

Large Language Models (LLMs) have become essential for interactive AI systems, yet they remain fundamentally static after deployment: they cannot update their parameters from interaction feedback and often repeat the same mistakes across long interaction streams. We propose Dual-Process Agent (DPA), a framework for continual context refinement that enables learning without modifying a frozen model backbone. Inspired by dual-process theory from cognitive science, DPA decomposes each interaction episode into two complementary processes: a fast System 1 that retrieves compact, relevant context from an explicit long-term memory and generates responses, and a slow System 2 that reflects on outcomes and writes curated updates back into memory. To prevent memory degradation over extended interactions, DPA maintains bulletized memory entries with utility statistics and employs a conservative curator gate that filters generic, redundant, or conflicting insertions. Experiments on six diverse benchmarks demonstrate that DPA consistently outperforms vanilla prompting and competitive baselines on both GPT-5.1 and Llama-3.1-8B backbones, achieving the best overall performance across multiple reasoning and knowledge-intensive tasks. Full article

Chronic lymphocytic leukemia (CLL) is uncommon in Asia, and longitudinal genomic data from Asian cohorts are limited. We conducted serial whole-exome sequencing (WES) in a multicenter Korean cohort of newly diagnosed, elderly CLL treated with chlorambucil–obinutuzumab to evaluate mutational heterogeneity and clonal hematopoiesis of indeterminate potential (CHIP) during treatment and follow-up. Tumor-only variants were filtered, restricted to nonsynonymous or loss-of-function coding/splice-site mutations, and summarized as a binary patient-by-gene matrix for principal component analysis (PCA), trajectory analysis, and k-means clustering. CHIP was defined as ≥1 qualifying mutation in a prespecified CHIP gene set. Baseline PCA was more compact in patients with complete response at end of treatment, whereas partial response or progressive disease cases were more dispersed. PCA trajectories were compact and directionally consistent in complete responders, more dispersed in partial responders, and highly heterogeneous without a dominant direction in progressive disease. Clustering identified dispersed and compact clusters, and CHIP-associated mutations were enriched in the dispersed cluster (55.6% vs. 8.3%, Fisher’s exact p = 0.0086). In paired samples collected 3–5 months after end of treatment, CHIP status changed in some patients. Serial WES may provide complementary information to treatment response, although these observations require confirmation in larger cohorts. Full article

In-depth research on marine biodiversity is essential for understanding and protecting marine ecosystems, where semantic segmentation of marine species plays a crucial role. However, segmenting microscopic zooplankton images remains challenging due to highly variable morphologies, complex boundaries, and the scarcity of high-quality pixel-level annotations that require expert knowledge. Existing semi-supervised methods often rely on single-model perspectives, producing unreliable pseudo-labels and limiting performance in such complex scenarios. To address these challenges, this paper proposes a consistency-driven dual-teacher framework tailored for zooplankton segmentation. Two heterogeneous teacher networks are employed: one captures global morphological features, while the other focuses on local fine-grained details, providing complementary and diverse supervision and alleviating overfitting under limited annotations. In addition, a dynamic fusion-based pseudo-label filtering strategy is introduced to adaptively integrate hard and soft labels by jointly considering prediction consistency and confidence scores, thereby enhancing supervision flexibility. Extensive experiments on the Zooplankton-21 Microscopic Segmentation Dataset (ZMS-21), a self-constructed microscopic zooplankton dataset demonstrate that the proposed method consistently outperforms existing semi-supervised segmentation approaches under various annotation ratios, achieving mIoU scores of 64.80%, 69.58%, 70.32%, and 73.92% with 1/16, 1/8, 1/4, and 1/2 labeled data, respectively. Full article