Search Results (123)

Aiming at the problems of complex image background and numerous interference factors faced by visual navigation systems in orchard environments, this paper proposes an orchard navigation line recognition method based on U-Net. Firstly, the drivable areas in the collected images are labeled using Labelme (a graphical tool for image annotation) to create an orchard dataset. Then, the Spatial Attention (SA) mechanism is inserted into the downsampling stage of the traditional U-Net semantic segmentation method, and the Coordinate Attention (CA) mechanism is added to the skip connection stage to obtain complete context information and optimize the feature restoration process of the drivable area in the field, thereby improving the overall segmentation accuracy of the model. Subsequently, the improved U-Net network is trained using the enhanced dataset to obtain the drivable area segmentation model. Based on the detected drivable area segmentation mask, the navigation line information is extracted, and the geometric center points are calculated row by row. After performing sliding window processing and bidirectional interpolation filling on the center points, the navigation line is generated through spline interpolation. Finally, the proposed method is compared and verified with U-Net, SegViT, SE-Net, and DeepLabv3+ networks. The results show that the improved drivable area segmentation model has a Recall of 90.23%, a Precision of 91.71%, a mean pixel accuracy (mPA) of 87.75%, and a mean intersection over union (mIoU) of 84.84%. Moreover, when comparing the recognized navigation line with the actual center line, the average distance error of the extracted navigation line is 56 mm, which can provide an effective reference for visual autonomous navigation in orchard environments. Full article

This paper presents a fault diagnosis model for rolling bearings that addresses the challenges of establishing long-sequence correlations and extracting spatial features in deep-learning models. The proposed model combines SENet with an improved Informer model. Initially, local features are extracted using the Conv1d method, and input data is optimized through normalization and embedding techniques. Next, the SE-Conv1d network model is employed to enhance key features while suppressing noise interference adaptively. In the improved Informer model, the ProbSparse self-attention mechanism and self-attention distillation technique efficiently capture global dependencies in long sequences within the rolling bearing dataset, significantly reducing computational complexity and improving accuracy. Finally, experiments on the CWRU and HUST datasets demonstrate that the proposed model achieves accuracy rates of 99.78% and 99.45%, respectively. The experimental results show that, compared to other deep learning methods, the proposed model offers superior fault diagnosis accuracy, stability, and generalization ability. Full article

Oceanic whitecaps are caused by wave breaking and are very important in air–sea interactions. Usually, whitecap coverage is considered a key factor in representing the role of whitecaps. However, the accurate identification of whitecap coverage in videos under dynamic marine conditions is a tough task. An EMA-SE-ResUNet deep learning model was proposed in this study to address this challenge. Based on a foundation of residual network (ResNet)-50 as the encoder and U-Net as the decoder, the model incorporated efficient multi-scale attention (EMA) module and squeeze-and-excitation network (SENet) module to improve its performance. By employing a dynamic weight allocation strategy and a channel attention mechanism, the model effectively strengthens the feature representation capability for whitecap edges while suppressing interference from wave textures and illumination noise. The model’s adaptability to complex sea surface scenarios was enhanced through the integration of data augmentation techniques and an optimized joint loss function. By applying the proposed model to a dataset collected by a shipborne camera system deployed during a comprehensive fishery resource survey in the northwest Pacific, the model results outperformed main segmentation algorithms, including U-Net, DeepLabv3+, HRNet, and PSPNet, in key metrics: whitecap intersection over union (IoU_W) = 73.32%, pixel absolute error (PAE) = 0.081%, and whitecap F1-score (F1_W) = 84.60. Compared to the traditional U-Net model, it achieved an absolute improvement of 2.1% in IoU_W while reducing computational load (GFLOPs) by 57.3% and achieving synergistic optimization of accuracy and real-time performance. This study can provide highly reliable technical support for studies on air–sea flux quantification and marine aerosol generation. Full article

Industrial bearing surface defect detection faces challenges such as complex image backgrounds, multi-scale defects, and insufficient feature extraction. To address these issues, this paper proposes an improved YOLOv8-GSS defect detection method. Initially, the network substitutes the standard convolution in the C2f module and Concat module within the neck module with lightweight convolutions, GsConv, thereby reducing computational costs. Subsequently, to better capture and represent crucial features in the images, an SENetV2 attention mechanism is integrated before the SPPF module at the backbone end, effectively enhancing the model’s accuracy and robustness in defect detection. Finally, a self-built dataset of surface images of bearing rings collected from industrial sites is utilized as the basis for extensive experimentation. Experimental results show that the network achieves 97.8% AP50, with detection accuracy for large-, medium-, and small-scale defects improved by 2.4%, 3.6%, and 2.3%, respectively.2.3% respectively. The detection speed reaches 115 frames per second (FPS). Compared to mainstream surface defect detection algorithms, the proposed method exhibits significant improvements in both accuracy and detection speed. Full article

Scene Text Recognition (STR) faces significant challenges under complex degradation conditions, such as distortion, occlusion, and semantic ambiguity. Most existing methods rely heavily on language priors for correction, but effectively constructing language rules remains a complex problem. This paper addresses two key challenges: (1) The over-correction behavior of language models, particularly on semantically deficient input, can result in both recognition errors and loss of critical information. (2) Character misalignment in visual features, which affects recognition accuracy. To address these problems, we propose a Deformable-Alignment-based Dual Correction Mechanism (DADCM) for STR. Our method includes the following key components: (1) We propose a visually guided and language-assisted correction strategy. A dynamic confidence threshold is used to control the degree of language model intervention. (2) We designed a visual backbone network called SCRTNet. The net enhances key text regions through a channel attention module (SENet) and applies deformable convolution (DCNv4) in deep layers to better model distorted or curved text. (3) We propose a deformable alignment module (DAM). The module combines Gumbel-Softmax-based anchor sampling and geometry-aware self-attention to improve character alignment. Experiments on multiple benchmark datasets demonstrate the superiority of our approach. Especially on the Union14M-Benchmark, where the recognition accuracy surpasses previous methods by 1.1%, 1.6%, 3.0%, and 1.3% on the Curved, Multi-Oriented, Contextless, and General subsets, respectively. Full article

Text-dependent speaker verification (TD-SV), which verifies speaker identity using predefined phrases, has gained attention as a reliable contactless biometric authentication method for smart devices, internet of things (IoT), and real-time applications. However, in real-world environments, limited training data, background noise, and microphone channel variability significantly degrade TD-SV performance, particularly on resource-constrained devices that require real-time inference. To address these challenges, we propose a lightweight end-to-end TD-SV model based on a convolution-augmented transformer (Conformer) architecture enhanced with a channel-temporal attention (CTA) module as an input enhancement that specifically targets speaker-discriminative patterns in short, fixed utterances. Unlike existing attention mechanisms (Squeeze-and-Excitation Networks (SENet), Convolutional Block Attention Module (CBAM)) designed for computer vision tasks, our CTA module employs frequency-wise statistical pooling to capture acoustic variability patterns crucial for speaker discrimination within identical phonetic content. Experiments conducted on the challenging far-field and noisy SLR 85 HI-MIA dataset demonstrate that the proposed CTA-Conformer achieves an equal error rate (EER) of 2.04% and a minimum detection cost function (minDCF) of 0.20, achieving competitive performance compared to recent TD-SV approaches. Additionally, INT8 quantization reduces the model size by 75.8%, significantly improves inference speed, and enabling real-time deployment on edge devices. Our approach thus offers a practical solution for robust and efficient TD-SV in embedded internet of things (IoT) environments. Full article

Spodoptera frugiperda is a major transboundary migratory pest under global alert by the Food and Agriculture Organization (FAO) of the United Nations. The accurate identification and counting of trapped adults in the field are key technologies for achieving quantitative monitoring and precision pest control. However, precise recognition is challenged by issues such as scale loss and the presence of mixed insect species in trapping images. To address this, we constructed a field image dataset of trapped Spodoptera frugiperda adults and proposed an improved YOLOv5s-based detection method. The dataset was collected over a two-year sex pheromone monitoring campaign in eastern–central Yunnan, China, comprising 9550 labeled insects across six categories, and was split into training, validation, and test sets in an 8:1:1 ratio. In this study, YOLOv7, YOLOv8, Mask R-CNN, and DETR were selected as comparative baselines to evaluate the recognition of images containing Spodoptera frugiperda adults and other insect species. However, the complex backgrounds introduced by field trap photography adversely affected classification performance, resulting in a relatively modest average accuracy. Considering the additional requirement for model lightweighting, we further enhanced the YOLOv5s architecture by integrating Mosaic data augmentation and an adaptive anchor box strategy. Additionally, three attention mechanisms—SENet, CBAM, and Coordinate Attention (CA)—were embedded into the backbone to build a multidimensional attention comparison framework, demonstrating CBAM’s superiority under complex backgrounds. Ultimately, the CBAM-YOLOv5 model achieved 97.8% mAP@0.5 for Spodoptera frugiperda identification, with recognition accuracy for other insect species no less than 72.4%. Based on the optimized model, we developed an intelligent recognition system capable of image acquisition, identification, and counting, offering a high-precision algorithmic solution for smart trapping devices. Full article

Bolted connections are extensively utilized in aerospace, civil, and mechanical systems for structural assembly. However, inevitable structural vibrations can induce bolt loosening, leading to preload reduction and potential structural failure. Early-stage preload degradation, particularly during initial loosening, is often undetectable by conventional monitoring methods due to limited sensitivity and poor noise resilience. To address these limitations, this study proposes an intelligent bolt preload monitoring framework that combines electromechanical impedance (EMI) signal analysis with a parallel deep learning architecture. A multiphysics-coupled model of flange joint connections is developed to reveal the nonlinear relationships between preload degradation and changes in EMI conductance spectra, specifically resonance peak shifts and amplitude attenuation. Based on this insight, a parallel convolutional neural network (P-CNN) is designed, employing dual branches with 1 × 3 and 1 × 7 convolutional kernels to extract local and global spectral features, respectively. The architecture integrates dilated convolution to expand frequency–domain receptive fields and an enhanced SENet-based channel attention mechanism to adaptively highlight informative frequency bands. Experimental validation on a flange-bolt platform demonstrates that the proposed P-CNN achieves 99.86% classification accuracy, outperforming traditional CNNs by 20.65%. Moreover, the model maintains over 95% accuracy with only 25% of the original training samples, confirming its robustness and data efficiency. The results demonstrate the feasibility and scalability of the proposed approach for real-time, small-sample, and noise-resilient structural health monitoring of bolted connections. Full article

Long-term and accurate prediction of air pollutant concentrations can serve as a foundation for air pollution warning and prevention, which is crucial for social development and human health. In this study, we provide a model for predicting the concentration of air pollutants based on big data spatiotemporal correlation analysis and deep learning methods. Based on an improved temporal convolutional network (TCN) and a bi-directional gated recurrent unit (BiGRU) as the fundamental architecture, the model adds two attention mechanisms to improve performance: Squeeze and Excitation Networks (SENet) and Convolutional Block Attention Module (CBAM). The improved TCN moves the residual connection layer to the network’s front end as a preprocessing procedure, improving the model’s performance and operating efficiency, particularly for big data jobs like air pollution concentration prediction. The use of SENet improves the model’s comprehension and extraction of long-term dependent features from pollutants and meteorological data. The incorporation of CBAM enhances the model’s perception ability towards key local regions through an attention mechanism in the spatial dimension of the feature map. The TCN-SENet-BiGRU-CBAM model successfully realizes the prediction of air pollutant concentrations by extracting the spatiotemporal features of the data. Compared with previous advanced deep learning models, the model has higher prediction accuracy and generalization ability. The model is suitable for prediction tasks from 1 to 12 h in the future, with root mean square error (RMSE) and mean absolute error (MAE) ranging from 5.309~14.043 and 3.507~9.200, respectively. Full article

With the development of smart agriculture, the precise identification of fruit tree trunks by orchard management robots has become a key technology for achieving autonomous navigation. To solve the issue of tree trunks being hard to see against their background in orchards, this study introduces PCC-YOLO (PENet, CoT-Net, and Coord-SE attention-based YOLOv8), a new trunk detection model based on YOLOv8. It improves the ability to identify features in low-contrast situations by using a pyramid enhancement network (PENet), a context transformer (CoT-Net) module, and a combined coordinate and channel attention mechanism. By introducing a pyramid enhancement network (PENet) into YOLOv8, the model’s feature extraction ability under low-contrast conditions is enhanced. A context transformer module (CoT-Net) is then used to strengthen global perception capabilities, and a combination of coordinate attention (Coord-Att) and SENetV2 is employed to optimize target localization accuracy. Experimental results show that PCC-YOLO achieves a mean average precision (mAP) of 82.6% on a self-built orchard dataset (5000 images) and a detection speed of 143.36 FPS, marking a 4.8% improvement over the performance of the baseline YOLOv8 model, while maintaining a low computational load (7.8 GFLOPs). The model demonstrates a superior balance of accuracy, speed, and computational cost compared to results for the baseline YOLOv8 and other common YOLO variants, offering an efficient solution for the real-time autonomous navigation of orchard management robots. Full article

Lung cancer is one of the leading causes of cancer-related mortality worldwide. The diagnosis of this disease remains a challenge due to the subtle and ambiguous nature of early-stage symptoms and imaging findings. Deep learning approaches, specifically Convolutional Neural Networks (CNNs), have significantly advanced medical image analysis. However, conventional architectures such as ResNet50 that rely on first-order pooling often fall short. This study aims to overcome the limitations of CNNs in lung cancer classification by proposing a novel and dynamic model named LungSE-SOP. The model is based on Second-Order Pooling (SOP) and Squeeze-and-Excitation Networks (SENet) within a ResNet50 backbone to improve feature representation and class separation. A novel Dynamic Feature Enhancement (DFE) module is also introduced, which dynamically adjusts the flow of information through SOP and SENet blocks based on learned importance scores. The model was trained using a publicly available IQ-OTH/NCCD lung cancer dataset. The performance of the model was assessed using various metrics, including the accuracy, precision, recall, F1-score, ROC curves, and confidence intervals. For multiclass tumor classification, our model achieved 98.6% accuracy for benign, 98.7% for malignant, and 99.9% for normal cases. Corresponding F1-scores were 99.2%, 99.8%, and 99.9%, respectively, reflecting the model’s high precision and recall across all tumor types and its strong potential for clinical deployment. Full article

Using Photoplethysmogram (PPG) signals for identity recognition has been proven effective in biometric authentication. However, in real-world applications, PPG signals are prone to interference from noise, physical activity, diseases, and other factors, making it challenging to ensure accurate user recognition and verification in complex environments. To address these issues, this paper proposes an improved MSF-SE ResNet50 (Multi-Scale Feature Squeeze-and-Excitation ResNet50) model based on 2D PPG signals. Unlike most existing methods that directly process one-dimensional PPG signals, this paper adopts a novel approach based on two-dimensional PPG signal processing. By applying Continuous Wavelet Transform (CWT), the preprocessed one-dimensional PPG signal is transformed into a two-dimensional time-frequency map, which not only preserves the time-frequency characteristics of the signal but also provides richer spatial information. During the feature extraction process, the SENet module is first introduced to enhance the ability to extract distinctive features. Next, a novel Lightweight Multi-Scale Feature Fusion (LMSFF) module is proposed, which addresses the limitation of single-scale feature extraction in existing methods by employing parallel multi-scale convolutional operations. Finally, cross-stage feature fusion is implemented, overcoming the limitations of traditional feature fusion methods. These techniques work synergistically to improve the model’s performance. On the BIDMC dataset, the MSF-SE ResNet50 model achieved accuracy, precision, recall, and F1 scores of 98.41%, 98.19%, 98.27%, and 98.23%, respectively. Compared to existing state-of-the-art methods, the proposed model demonstrates significant improvements across all evaluation metrics, highlighting its significance in terms of network architecture and performance. Full article

Traditional vehicle emergency braking research suffers from inaccurate maximum road adhesion coefficient identification and suboptimal wheel slip ratio control. To address these challenges in electronic hydraulic braking systems’ anti-lock braking technology, firstly, this paper proposes a CDOA-SENet-CNN neural network to precisely estimate the maximum road adhesion coefficient by monitoring and analyzing the braking process. Secondly, correlation curves between peak adhesion coefficients and ideal slip ratios are established using the Burckhardt model and CarSim 2020, and the estimated maximum adhesion coefficient from the CDOA-SENet-CNN network is used with these curves to determine the optimal slip ratio for the single-neuron integral sliding mode control (SNISMC) algorithm. Finally, an SNISMC control strategy is developed to adjust the wheel slip ratio to the optimal value, achieving stable wheel control across diverse road surfaces. Results indicate that the CDOA-SENet-CNN network rapidly and accurately estimates the peak braking surface adhesion coefficient. The SNISMC control strategy significantly enhances wheel slip ratio control, consequently increasing the effectiveness of vehicle brakes. This paper introduces an innovative, stable, and efficient solution for enhancing vehicle braking safety. Full article

To address the challenge of fault diagnosis for in-wheel motors in four-wheel independent driving systems under variable driving conditions and harsh environments, this paper proposes a novel method based on two-stream 2DCNNs (two-dimensional convolutional neural networks) with a DCBA (depthwise convolution block attention) module. The main contributions are twofold: (1) A DCBA module is introduced to extract multi-scale features—including prominent, local, and average information—from grayscale images reconstructed from vibration signals across different domains; and (2) a two-stream network architecture is designed to learn complementary feature representations from time-domain and time–frequency-domain signals, which are fused through fully connected layers to improve diagnostic accuracy. Experimental results demonstrate that the proposed method achieves high recognition accuracy under various working speeds, loads, and road surfaces. Comparative studies with SENet, ECANet, CBAM, and single-stream 2DCNN models confirm its superior performance and robustness. The integration of DCBA with dual-domain feature learning effectively enhances fault feature extraction under complex operating conditions. Full article

Accurate and timely forecasting of cross-regional tourist flows is essential for sustainable destination management, yet existing models struggle with sparse data, complex spatiotemporal interactions, and limited interpretability. This paper presents SE-TFF, a multi-scale tourism-flow forecasting framework that couples a Squeeze-and-Excitation (SE) network with reinforcement-driven optimization to adaptively re-weight environmental, economic, and social features. A benchmark dataset of 17.8 million records from 64 countries and 743 cities (2016–2024) is compiled from the Open Travel Data repository in github (OPTD) for training and validation. SE-TFF introduces (i) a multi-channel SE module for fine-grained feature selection under heterogeneous conditions, (ii) a Top-K attention filter to preserve salient context in highly sparse matrices, and (iii) a Double-DQN layer that dynamically balances prediction objectives. Experimental results show SE-TFF attains 56.5% MAE and 65.6% RMSE reductions over the best baseline (ARIMAX) at 20% sparsity, with 0.92 × 10³ average MAE across multi-task outputs. SHAP analysis ranks climate anomalies, tourism revenue, and employment as dominant predictors. These gains demonstrate SE-TFF’s ability to deliver real-time, interpretable forecasts for data-limited destinations. Future work will incorporate real-time social media signals and larger multimodal datasets to enhance generalizability. Full article