Search Results (804)

Environmental perception is an essential task for autonomous driving, which is typically based on LiDAR or camera sensors. In recent years, 4D mm-Wave radar, which acquires 3D point cloud together with point-wise Doppler velocities, has drawn substantial attention owing to its robust performance under adverse weather conditions. Nonetheless, due to the high sparsity and substantial noise inherent in radar measurements, most radar perception studies are limited to object-level tasks, with point-level tasks such as semantic segmentation remaining largely underexplored. This paper aims to explore the possibility of using 4D radar in semantic segmentation. We set up the ZJUSSet dataset containing accurate point-wise class labels for radar and LiDAR. Then we propose a cross-modal distillation framework RaSS to fulfill the task. An adaptive Doppler compensation module is also designed to facilitate the segmentation. Experimental results on ZJUSSet and VoD dataset demonstrate that our RaSS model significantly outperforms the baselines and competitors. Code and dataset will be available upon paper acceptance. Full article

With the rapid expansion of multi-platform remote sensing applications, cloud contamination significantly impedes cross-platform data utilization. Current cloud detection methods face critical technical challenges in cross-platform settings, including neglect of atmospheric radiative transfer mechanisms, inadequate multi-scale structural decoupling, high intra-class variability coupled with inter-class similarity, cloud boundary ambiguity, cross-modal feature inconsistency, and noise propagation in pseudo-labels within semi-supervised frameworks. To address these issues, we introduce a Physics-Guided Multi-Representation Network (PGMRN) that adopts a student–teacher architecture and fuses tri-modal representations—Pseudo-NDVI, structural, and textural features—via atmospheric priors and intrinsic image decomposition. Specifically, PGMRN first incorporates an InfoNCE contrastive loss to enhance intra-class compactness and inter-class discrimination while preserving physical consistency; subsequently, a boundary-aware regional adaptive weighted cross-entropy loss integrates PA-CAM confidence with distance transforms to refine edge accuracy; furthermore, an Uncertainty-Aware Quadruple Consistency Propagation (UAQCP) enforces alignment across structural, textural, RGB, and physical modalities; and finally, a dynamic confidence-screening mechanism that couples PA-CAM with information entropy and percentile-based thresholding robustly refines pseudo-labels. Extensive experiments on four benchmark datasets demonstrate that PGMRN achieves state-of-the-art performance, with Mean IoU values of 70.8% on TCDD, 79.0% on HRC_WHU, and 83.8% on SWIMSEG, outperforming existing methods. Full article

In recent years, solar energy has emerged as a pillar of sustainable development. However, maintaining panel efficiency under extreme environmental conditions remains a persistent hurdle. This study introduces an automated defect detection pipeline that leverages deep learning and computer vision to identify five standard anomaly classes: Non-Defective, Dust, Defective, Physical Damage, and Snow on photovoltaic surfaces. To build a robust foundation, a heterogeneous dataset of 8973 images was sourced from public repositories and standardized into a uniform labeling scheme. This dataset was then expanded through an aggressive augmentation strategy, including flips, rotations, zooms, and noise injections. A YOLOv11-based model was trained and fine-tuned using both fixed and adaptive learning rate schedules, achieving a mAP@0.5 of 85% and accuracy, recall, and F1-score above 95% when evaluated across diverse lighting and dust scenarios. The optimized model is integrated into an interactive dashboard that processes live camera streams, issues real-time alerts upon defect detection, and supports proactive maintenance scheduling. Comparative evaluations highlight the superiority of this approach over manual inspections and earlier YOLO versions in both precision and inference speed, making it well suited for deployment on edge devices. Automating visual inspection not only reduces labor costs and operational downtime but also enhances the longevity of solar installations. By offering a scalable solution for continuous monitoring, this work contributes to improving the reliability and cost-effectiveness of large-scale solar energy systems. Full article

Control point detection is a critical initial step in camera calibration. For checkerboard corner points, detection is based on inferences about local gradients in the image. Infrared (IR) imaging, however, poses challenges due to its low resolution and low signal-to-noise ratio, hindering the identification of clear local features. This study proposes a physics-informed neural network (PINN) based on the YOLO target detection model to detect checkerboard corner points in infrared images, aiming to enhance the calibration accuracy of infrared thermal cameras. This method first optimizes the YOLO model used for corner detection based on the idea of enhancing image gradient information extraction and then incorporates camera physical information into the training process so that the model can learn the intrinsic constraints between corner coordinates. Camera physical information is applied to the loss calculation process during training, avoiding the impact of label errors on the model and further improving detection accuracy. Compared with the baselines, the proposed method reduces the root mean square error (RMSE) by at least 30% on average across five test sets, indicating that the PINN-based corner detection method can effectively handle low-quality infrared images and achieve more accurate camera calibration. Full article

Early detection of physiological dysregulation is critical for timely intervention and effective health management. Traditional monitoring systems often rely on labeled data and predefined thresholds, limiting their adaptability and generalization to unseen conditions. To address this, we propose a framework for label-free classification of physiological states using Heart Rate Variability (HRV), combined with unsupervised machine learning techniques. This approach is particularly valuable when annotated datasets are scarce or unavailable—as is often the case in real-world wearable and IoT-based health monitoring. In this study, data were collected from participants under controlled conditions representing rest, stress, and physical exertion. Core HRV parameters such as the SDNN (Standard Deviation of all Normal-to-Normal intervals), RMSSD (Root Mean Square of the Successive Differences), DFA (Detrended Fluctuation Analysis) were extracted. Principal Component Analysis was applied for dimensionality reduction. K-Means, hierarchical clustering, and Density-based spatial clustering of applications with noise (DBSCAN) were used to uncover natural groupings within the data. DBSCAN identified outliers associated with atypical responses, suggesting potential for early anomaly detection. The combination of HRV descriptors enabled unsupervised classification with over 90% consistency between clusters and physiological conditions. The proposed approach successfully differentiated the three physiological conditions based on HRV and fractal features, with a clear separation between clusters in terms of DFA α1, α2, LF/HF, and RMSSD (with high agreement to physiological labels (Purity ≈ 0.93; ARI = 0.89; NMI = 0.92)). Furthermore, DBSCAN identified three outliers with atypical autonomic profiles, highlighting the potential of the method for early warning detection in real-time monitoring systems. Full article

Sweetpotato virus disease (SPVD) poses a significant threat to global sweetpotato production; therefore, early, accurate field-scale detection is necessary. To address the limitations of the currently utilized assays, we propose PLCNet (Plant-Level Classification Network), a rapid, non-destructive SPVD identification framework using UAV-acquired hyperspectral imagery. High-resolution data from early sweetpotato growth stages were processed via three feature selection methods—Random Forest (RF), Minimum Redundancy Maximum Relevance (mRMR), and Local Covariance Matrix (LCM)—in combination with 24 vegetation indices. Variance Inflation Factor (VIF) analysis reduced multicollinearity, yielding an optimized SPVD-sensitive feature set. First, using the RF-selected bands and vegetation indices, we benchmarked four classifiers—Support Vector Machine (SVM), Gradient Boosting Decision Tree (GBDT), Residual Network (ResNet), and 3D Convolutional Neural Network (3D-CNN). Under identical inputs, the 3D-CNN achieved superior performance (OA = 96.55%, Macro F1 = 95.36%, UA_mean = 0.9498, PA_mean = 0.9504), outperforming SVM, GBDT, and ResNet. Second, with the same spectral–spatial features and 3D-CNN backbone, we compared a pixel-level baseline (CropdocNet) against our plant-level PLCNet. CropdocNet exhibited spatial fragmentation and isolated errors, whereas PLCNet’s two-stage pipeline—deep feature extraction followed by connected-component analysis and majority voting—aggregated voxel predictions into coherent whole-plant labels, substantially reducing noise and enhancing biological interpretability. By integrating optimized feature selection, deep learning, and plant-level post-processing, PLCNet delivers a scalable, high-throughput solution for precise SPVD monitoring in agricultural fields. Full article

Impulsive noise suppression plays a crucial role in enhancing the reliability of power line communication (PLC). In view of the rapid advancement of deep learning methodologies, recently, studies on deep learning-based impulsive noise suppression have garnered extensive attention. Nevertheless, on one hand, the training of deep learning-based impulsive noise suppression models relies on a large number of labeled data, whose acquisition incurs high costs. On the other hand, the currently proposed models struggle to adapt to the dynamic variations in impulsive noise distributions. To address these two issues, in this paper, a UNet-based self-supervised learning model for impulsive noise suppression (UNet-INSN) is proposed. Firstly, by using the designed global mask mapper, UNet-INSN can utilize the entire noisy signal for model training, resolving the information loss issue caused by partial signal masking in traditional mask-driven algorithms. Secondly, a reproducibility loss function is introduced to effectively prevent the model from degenerating into an identity mapping, thereby enhancing the denoising performance of UNet-INSN. Simulation results show that the required SNRs for the proposed algorithm to achieve a bit error rate of 10⁻⁶ under ideal and non-ideal conditions are 12 dB and 26 dB, respectively, significantly outperforming comparison methods. Moreover, it still exhibits excellent robustness and generalization capabilities when the impulsive noise distribution changes dynamically. Full article

To address the challenges of signal recognition and prediction in intelligent sliding sleeve downlink communication systems, this paper proposes a dual-model framework based on Long Short-Term Memory (LSTM) networks. The system comprises a classifier for identifying pressure wave edge types and a generator for predicting pressure waveforms. High-quality training data are generated by simulating pressure wave propagation caused by throttle valve modulations. A sliding window strategy and Z-score normalization are applied to enhance temporal modeling. The classifier achieves a high accuracy in identifying rising and falling edges under noise-free conditions. The generator, trained on down-sampled waveform segments, accurately reconstructs pressure dynamics using a dual-input strategy based on historical segments and hypothetical labels. A residual-based decision mechanism is employed to complete the full sequence label prediction. To evaluate robustness, noise intensities of 30 dB and 40 dB are introduced. The proposed system maintains high performance under both conditions, achieving label prediction accuracies of 100%. Error metrics such as MAE and RMSE remain within acceptable bounds, even in noisy environments. The results demonstrate that the proposed LSTM-based method has been validated on simulated data, showing its potential to approximate performance in real-world conditions. It provides a promising solution for cable-free measurement-while-drilling (MWD) and remote control of intelligent sliding sleeves in complex downhole environments. Full article

Synthetic aperture radar (SAR) image classification under limited data conditions faces two major challenges: inter-class similarity, where distinct radar targets (e.g., tanks and armored trucks) have nearly identical scattering characteristics, and intra-class variability, caused by speckle noise, pose changes, and differences in depression angle. To address these challenges, we propose MHD-ProtoNet, a meta-learning framework that extends prototypical networks with two key innovations: margin-aware hard example mining to better separate confusable classes by enforcing prototype distance margins, and dual-loss optimization to refine embeddings and improve robustness to noise-induced variations. Evaluated on the MSTAR dataset in a five-way one-shot task, MHD-ProtoNet achieves $76.80\%$ accuracy, outperforming the Hybrid Inference Network (HIN) $(74.70\%)$, as well as standard few-shot methods such as prototypical networks $(69.38\%)$, ST-PN $(72.54\%)$, and graph-based models like ADMM-GCN $(61.79\%)$ and DGP-NET $(68.60\%)$. By explicitly mitigating inter-class ambiguity and intra-class noise, the proposed model enables robust SAR target recognition with minimal labeled data. Full article

In hyperspectral image (HSI) classification, feature learning and label accuracy play a crucial role. In actual hyperspectral scenes, however, noisy labels are unavoidable and seriously impact the performance of methods. While deep learning has achieved remarkable results in HSI classification tasks, its noise-resistant performance usually comes at the cost of feature representation capabilities. High-dimensional and deep convolution can capture rich deep semantic features, but with high complexity and resource consumption. To deal with these problems, we propose a CViT Weakly Supervised Network (CWSN) for HSI classification. Specifically, a lightweight 1D-2D two-branch network is used for local generalization and enhancement of spatial–spectral features. Then, the fusion and characterization of local and global features are achieved through the CNN-Vision Transformer (CViT) cascade strategy. The experimental results on four benchmark HSI datasets show that CWSN has good anti-noise ability and ensures the robustness and versatility of the network facing both clean and noisy training sets. Compared to other methods, the CWSN has better classification accuracy. Full article

In order to solve the problems of modulation signals in low signal-to-noise ratio (SNR), such as poor feature extraction ability, strong dependence on single modal data, and insufficient recognition accuracy, this paper proposes a multi-dimensional feature MFCA-transformer recognition network that integrates phase, frequency and power information. The network uses Triple Dynamic Feature Fusion (TDFF) to fuse constellation, time-frequency, and power spectrum features through the adaptive dynamic mechanism to improve the quality of feature fusion. A Channel Prior Convolutional Attention (CPCA) module is introduced to solve the problem of insufficient information interaction between different channels in multi-dimensional feature recognition tasks, promote information transmission between various feature channels, and enhance the recognition ability of the model for complex features. The label smoothing technique is added to the loss function to reduce the overfitting of the model to the specific label and improve the generalization ability of the model by adjusting the distribution of the real label. Experiments show that the recognition accuracy of the proposed method is significantly improved on the public datasets, at high signal-to-noise ratios, the recognition accuracy can reach 93.2%, which is 3% to 14% higher than those of the existing deep learning recognition methods. Full article

Domain adaptation is a key approach to ensure that artificial intelligence models maintain reliable performance when facing distributional shifts between training (source) and testing (target) domains. However, existing methods often struggle to simultaneously preserve domain-invariant representations and discriminative class structures, particularly in the presence of complex covariate shifts and noisy pseudo-labels in the target domain. In this work, we introduce Conditional Rényi $\alpha$-Entropy Domain Adaptation, named CREDA, a novel deep learning framework for domain adaptation that integrates kernel-based conditional alignment with a differentiable, matrix-based formulation of Rényi’s quadratic entropy. The proposed method comprises three main components: (i) a deep feature extractor that learns domain-invariant representations from labeled source and unlabeled target data; (ii) an entropy-weighted approach that down-weights low-confidence pseudo-labels, enhancing stability in uncertain regions; and (iii) a class-conditional alignment loss, formulated as a Rényi-based entropy kernel estimator, that enforces semantic consistency in the latent space. We validate CREDA on standard benchmark datasets for image classification, including Digits, ImageCLEF-DA, and Office-31, showing competitive performance against both classical and deep learning-based approaches. Furthermore, we employ nonlinear dimensionality reduction and class activation maps visualizations to provide interpretability, revealing meaningful alignment in feature space and offering insights into the relevance of individual samples and attributes. Experimental results confirm that CREDA improves cross-domain generalization while promoting accuracy, robustness, and interpretability. Full article

Deep-learning-based multiple label chest X-ray classification has achieved significant success, but existing models still have three main issues: fixed-scale convolutions fail to capture both large and small lesions, standard pooling is lacking in the lack of attention to important regions, and linear classification lacks the capacity to model complex dependency between features. To circumvent these obstacles, we propose CONVFCMAE, a lightweight yet powerful framework that is built on a backbone that is partially frozen (77.08 % of the initial layers are fixed) in order to preserve complex, multi-scale features while decreasing the number of trainable parameters. Our architecture adds (1) an intelligent global pooling module that is learnable, with $1\times 1$ convolutions that are dynamically weighted by their spatial location, and (2) a multi-head attention block that is dedicated to channel re-calibration, along with (3) a two-layer MLP that has been enhanced with ReLU, batch normalization, and dropout. This module is used to enhance the non-linearity of the feature space. To further reduce the noise associated with labels and the imbalance in class distribution inherent to the NIH ChestXray14 dataset, we utilize a combined loss that combines BCEWithLogits and Focal Loss as well as extensive data augmentation. On ChestXray14, the average ROC–AUC of CONVFCMAE is 0.852, which is 3.97 percent greater than the state of the art. Ablation experiments demonstrate the individual and collective effectiveness of each component. Grad-CAM visualizations have a superior capacity to localize the pathological regions, and this increases the interpretability of the model. Overall, CONVFCMAE provides a practical, generalizable solution to the problem of extracting features from medical images in a practical manner. Full article

Ultrasound array imaging frequently employs a coupling medium to facilitate wave transmission from the transducer to the target component. Surface echoes, identified by their high-amplitude peaks, are crucial for determining the Time of Flight (TOF) in each channel, which is essential for deriving imaging focal laws. Accurate TOF measurement is vital in numerous applications, such as Non-Destructive Testing (NDT) and medical imaging. Conventional methods, such as threshold crossing and peak search, are highly sensitive to noise and spurious signals, therefore, more robust estimation techniques are needed. This study explores the application of a deep 3D Convolutional Neural Network (CNN) to detect surface echoes in Full Matrix Capture (FMC) ultrasound data. The CNN was trained on signals obtained with a matrix array and a set of reference components, utilizing a robotic arm setup to ensure precise probe positioning. Theoretical TOFs were computed based on the setup geometry to generate labeled training data. Test results indicated that the CNN model, which we have called DeepEcho3D, closely aligned with the ground truth and significantly reduced TOF estimation outliers (up to 98%) compared to traditional methods, demonstrating its potential for improved accuracy in surface echo detection. Full article

This article presents a novel method for detecting series arc faults (SAFs) in residential applications using time-domain imaging (TDI) and Long Short-Term Memory (LSTM) networks. The proposed method transforms current signals into grayscale images by filtering out the fundamental frequency, allowing key arc fault characteristics—such as high-frequency noise and waveform distortions—to become visually apparent. The use of Ensemble Empirical Mode Decomposition (EEMD) helped isolate meaningful signal components, although it was computationally intensive. To address real-time requirements, a simpler yet effective TDI method was developed for generating 2D images from current data. These images were then used as inputs to an LSTM network, which captures temporal dependencies and classifies both arc faults and appliance types. The proposed TDI-LSTM model was trained and tested on 7000 labeled datasets across five common household appliances. The experimental results show an average detection accuracy of 98.1%, with reduced accuracy for loads using thyristors (e.g., dimmers). The method is robust across different appliance types and conditions; comparisons with prior methods indicate that the proposed TDI-LSTM approach offers superior accuracy and broader applicability. Trade-offs in sampling rates and hardware implementation were discussed to balance accuracy and system cost. Overall, the TDI-LSTM approach offers a highly accurate, efficient, and scalable solution for series arc fault detection in smart home systems. Full article