Search Results (1,855)

This study proposes a novel rolling bearing fault diagnosis technique based on a synchrosqueezing wavelet transform (SWT) and a transfer residual convolutional neural network (TRCNN) designed to address the difficulties of feature extraction caused by the non-stationarity of fault signals, as well as the issue of low fault diagnosis accuracy resulting from small sample quantities. This approach transforms the one-dimensional vibration signal into time–frequency diagrams using an SWT based on complex Morlet wavelet basis functions, which redistributes (squeezes) the values of the wavelet coefficients at different localized points in a time–frequency plane to the estimated instantaneous frequencies. This allows the energy to be more fully concentrated in actual corresponding frequency components. This strategy improves both the time–frequency aggregation and the resolution, which better reflects the eigenvalues of non-stationary signals. In this process, transfer learning and a residual structure are used in the training of a convolutional neural network. The resulting time–frequency diagrams, acquired using the steps discussed above, are then input to the TRCNN for diagnosis. A series of validation experiments confirmed that applying the TRCNN structure made it possible to achieve high diagnostic accuracy, even when training the network with only a small number of fault samples, as all 12 fault types from the test dataset were diagnosed correctly. Further simulation experiments demonstrated that our proposed method improved fault diagnosis accuracy compared to that of conventional techniques (with increases of 1.74% over RCNN, 1.28% over TCNN, 1.62% over STFT, 1.73% over WT, 2.83% over PWVD, and 1.39% over STFA-PD). In addition, diagnostic accuracy reached 100% during the application of three-time transfer learning, validating the effectiveness of the proposed method for rolling bearing fault diagnosis. Full article

Convolutional neural networks have achieved excellent results in image denoising; however, there are still some problems: (1) The majority of single-branch models cannot fully exploit the image features and often suffer from the loss of information. (2) Most of the deep CNNs have inadequate edge feature extraction and saturated performance problems. To solve these problems, this paper proposes a two-branch convolutional image denoising network based on nonparametric attention and multiscale feature fusion, aiming to improve the denoising performance while better recovering the image edge and texture information. Firstly, ordinary convolutional layers were used to extract shallow features of noise in the image. Then, a combination of two-branch networks with different and complementary structures was used to extract deep features from the noise information in the image to solve the problem of insufficient feature extraction by the single-branch network model. The upper branch network used densely connected blocks to extract local features of the noise in the image. The lower branch network used multiple dilation convolution residual blocks with different dilation rates to increase the receptive field and extend more contextual information to obtain the global features of the noise in the image. It not only solved the problem of insufficient edge feature extraction but also solved the problem of the saturation of deep CNN performance. In this paper, a nonparametric attention mechanism is introduced in the two-branch feature extraction module, which enabled the network to pay attention to and learn the key information in the feature map, and improved the learning performance of the network. The enhanced features were then processed through the multiscale feature fusion module to obtain multiscale image feature information at different depths to obtain more robust fused features. Finally, the shallow features and deep features were summed using a long jump join and were processed through an ordinary convolutional layer and output to obtain a residual image. In this paper, Set12, BSD68, Set5, CBSD68, and SIDD are used as a test dataset to which different intensities of Gaussian white noise were added for testing and compared with several mainstream denoising methods currently available. The experimental results showed that this paper’s algorithm had better objective indexes on all test sets and outperformed the comparison algorithms. The method in this paper not only achieved a good denoising effect but also effectively retained the edge and texture information of the original image. The proposed method provided a new idea for the study of deep neural networks in the field of image denoising. Full article

Wheat is a globally cultivated cereal crop with substantial protein content present in its seeds. This research aimed to develop robust methods for predicting seed protein concentration in wheat seeds using bench-top hyperspectral imaging in the visible, near-infrared (VNIR), and shortwave infrared (SWIR) regions. To fully utilize the spectral and texture features of the full VNIR and SWIR spectral domains, a computer-vision-aided image co-registration methodology was implemented to seamlessly align the VNIR and SWIR bands. Sensitivity analyses were also conducted to identify the most sensitive bands for seed protein estimation. Convolutional neural networks (CNNs) with attention mechanisms were proposed along with traditional machine learning models based on feature engineering including Random Forest (RF) and Support Vector Machine (SVM) regression for comparative analysis. Additionally, the CNN classification approach was used to estimate low, medium, and high protein concentrations because this type of classification is more applicable for breeding efforts. Our results showed that the proposed CNN with attention mechanisms predicted wheat protein content with R² values of 0.70 and 0.65 for ventral and dorsal seed orientations, respectively. Although, the R² of the CNN approach was lower than of the best performing feature-based method, RF (R² of 0.77), end-to-end prediction capabilities with CNN hold great promise for the automation of wheat protein estimation for breeding. The CNN model achieved better classification of protein concentrations between low, medium, and high protein contents, with an R² of 0.82. This study’s findings highlight the significant potential of hyperspectral imaging and machine learning techniques for advancing precision breeding practices, optimizing seed sorting processes, and enabling targeted agricultural input applications. Full article

With the rapid development of the internet, network security threats are becoming increasingly complex and diverse, making traditional intrusion detection systems (IDSs) inadequate for handling the growing variety of sophisticated attacks. In particular, traditional methods based on rule matching and manual feature extraction demonstrate significant limitations in dealing with small samples and unknown attacks. This paper proposes an intrusion detection system based on multi-level feature extraction and inductive learning (MFEI-IDS) to address these challenges. The model innovatively integrates Fully Convolutional Networks (FCNs) with the Transformer architecture (FCN–Transformer) for feature extraction and utilizes an inductive learning component for efficient classification. The FCN–Transformer Encoder extracts multi-level features from raw network traffic, capturing local spatial patterns and global temporal dependencies, significantly enhancing the representation of network traffic while reducing reliance on manual feature engineering. The inductive learning module employs a dynamic routing mechanism to map sample feature vectors into robust class vector representations, achieving superior generalization when detecting unseen attack types. Compared to existing FCN–Transformer models, MFEI-IDS incorporates inductive learning to handle data imbalance and small-sample scenarios. Experiments on ISCX 2012 and CIC-IDS 2017 datasets show that MFEI-IDS outperforms mainstream IDS methods in accuracy, precision, recall, and F1-score, excelling in cross-dataset validation and demonstrating strong generalization capabilities. These results validate the practical potential of MFEI-IDS in small-sample learning, unknown attack detection, and dynamic network environments. Full article

The early and precise identification of a brain tumour is imperative for enhancing a patient’s life expectancy; this can be facilitated by quick and efficient tumour segmentation in medical imaging. Automatic brain tumour segmentation tools in computer vision have integrated powerful deep learning architectures to enable accurate tumour boundary delineation. Our study aims to demonstrate improved segmentation accuracy and higher statistical stability, using datasets obtained from diverse imaging acquisition parameters. This paper introduces a novel, fully automated model called Enhanced Channel Attention Transformer (E-CATBraTS) for Brain Tumour Semantic Segmentation; this model builds upon 3D CATBraTS, a vision transformer employed in magnetic resonance imaging (MRI) brain tumour segmentation tasks. E-CATBraTS integrates convolutional neural networks and Swin Transformer, incorporating channel shuffling and attention mechanisms to effectively segment brain tumours in multi-modal MRI. The model was evaluated on four datasets containing 3137 brain MRI scans. Through the adoption of E-CATBraTS, the accuracy of the results improved significantly on two datasets, outperforming the current state-of-the-art models by a mean DSC of 2.6% while maintaining a high accuracy that is comparable to the top-performing models on the other datasets. The results demonstrate that E-CATBraTS achieves both high segmentation accuracy and elevated generalisation abilities, ensuring the model is robust to dataset variation. Full article

In recent years, the fusion of hyperspectral and multispectral images in remote sensing image processing still faces challenges, primarily due to their complexity and multimodal characteristics. Diffusion models, known for their stable training process and exceptional image generation capabilities, have shown good application potential in this field. However, when dealing with multimodal data, it may prove challenging for the models to fully capture the intricate relationships between the modalities, which may result in incomplete information integration and a small amount of remaining noise in the generated images. To address these problems, we propose a new model, KanDiff, for hyperspectral and multispectral image fusion. To address the differences in modal information between multispectral and hyperspectral images, KANDiff incorporates Kolmogorov–Arnold Networks (KAN) to guide the inputs. It helps the model understand the complex relationships between the modalities by replacing the fixed activation function in the traditional MLP with a learnable activation function. Furthermore, the image generated by the diffusion model may exhibit a small amount of the remaining noise. Convolutional Neural Networks (CNNs) effectively extract local features through their convolutional layers and achieve noise suppression via layer-by-layer feature representation. Therefore, the MergeCNN module is further introduced to enhance the fusion effect, resulting in smoother and more accurate outcomes. Experimental results on the public CAVE and Harvard datasets indicate that KanDiff has achieved improvements over current high-performance methods across several metrics, particularly showing significant enhancements in the peak signal-to-noise ratio (PSNR) and the structural similarity index measure (SSIM), thus demonstrating superior performance. Additionally, we have created an image fusion dataset of the lunar surface, and KANDiff exhibits robust performance on this dataset as well. This work introduces an effective solution for addressing the challenges posed by missing high-resolution hyperspectral images (HRHS) data, which is essential for tasks including landing site selection and resource exploration within the realm of deep space exploration. Full article

Objectives: Accurate kidney and tumor segmentation of computed tomography (CT) scans is vital for diagnosis and treatment, but manual methods are time-consuming and inconsistent, highlighting the value of AI automation. This study develops a fully automated AI model using vision transformers (ViTs) and convolutional neural networks (CNNs) to detect and segment kidneys and kidney tumors in Contrast-Enhanced (CECT) scans, with a focus on improving sensitivity for small, indistinct tumors. Methods: The segmentation framework employs a ViT-based model for the kidney organ, followed by a 3D UNet model with enhanced connections and attention mechanisms for tumor detection and segmentation. Two CECT datasets were used: a public dataset (KiTS23: 489 scans) and a private institutional dataset (Private: 592 scans). The AI model was trained on 389 public scans, with validation performed on the remaining 100 scans and external validation performed on all 592 private scans. Tumors were categorized by TNM staging as small (≤4 cm) (KiTS23: 54%, Private: 41%), medium (>4 cm to ≤7 cm) (KiTS23: 24%, Private: 35%), and large (>7 cm) (KiTS23: 22%, Private: 24%) for detailed evaluation. Results: Kidney and kidney tumor segmentations were evaluated against manual annotations as the reference standard. The model achieved a Dice score of 0.97 ± 0.02 for kidney organ segmentation. For tumor detection and segmentation on the KiTS23 dataset, the sensitivities and average false-positive rates per patient were as follows: 0.90 and 0.23 for small tumors, 1.0 and 0.08 for medium tumors, and 0.96 and 0.04 for large tumors. The corresponding Dice scores were 0.84 ± 0.11, 0.89 ± 0.07, and 0.91 ± 0.06, respectively. External validation on the private data confirmed the model’s effectiveness, achieving the following sensitivities and average false-positive rates per patient: 0.89 and 0.15 for small tumors, 0.99 and 0.03 for medium tumors, and 1.0 and 0.01 for large tumors. The corresponding Dice scores were 0.84 ± 0.08, 0.89 ± 0.08, and 0.92 ± 0.06. Conclusions: The proposed model demonstrates consistent and robust performance in segmenting kidneys and kidney tumors of various sizes, with effective generalization to unseen data. This underscores the model’s significant potential for clinical integration, offering enhanced diagnostic precision and reliability in radiological assessments. Full article

Satellites frequently encounter atmospheric haze during imaging, leading to the loss of detailed information in remote sensing images and significantly compromising image quality. This detailed information is crucial for applications such as Earth observation and environmental monitoring. In response to the above issues, this paper proposes an end-to-end multi-scale adaptive feature extraction method for remote sensing image dehazing (MSD-Net). In our network model, we introduce a dilated convolution adaptive module to extract global and local detail features of remote sensing images. The design of this module can extract important image features at different scales. By expanding convolution, the receptive field is expanded to capture broader contextual information, thereby obtaining a more global feature representation. At the same time, a self-adaptive attention mechanism is also used, allowing the module to automatically adjust the size of its receptive field based on image content. In this way, important features suitable for different scales can be flexibly extracted to better adapt to the changes in details in remote sensing images. To fully utilize the features at different scales, we also adopted feature fusion technology. By fusing features from different scales and integrating information from different scales, more accurate and rich feature representations can be obtained. This process aids in retrieving lost detailed information from remote sensing images, thereby enhancing the overall image quality. A large number of experiments were conducted on the HRRSD and RICE datasets, and the results showed that our proposed method can better restore the original details and texture information of remote sensing images in the field of dehazing and is superior to current state-of-the-art methods. Full article

A type of aquaculture pond called a dike-pond system is distributed in the low-lying river delta of China’s eastern coast. Along with the swift growth of the coastal economy, the water surfaces of the dike-pond system (WDPS) play a major role attributed to pond aquaculture yielding more profits than dike agriculture. This study aims to explore the performance of deep learning methods for extracting WDPS from high spatial resolution remote sensing images. We developed three fully convolutional network (FCN) models: SegNet, UNet, and UNet++, which are compared with two traditional methods in the same testing regions from the Guangdong–Hong Kong–Macao Greater Bay Area. The extraction results of the five methods are evaluated in three parts. The first part is a general comparison that shows the biggest advantage of the FCN models over the traditional methods is the P-score, with an average lead of 13%, but the R-score is not ideal. Our analysis reveals that the low R-score problem is due to the omission of the outer ring of WDPS rather than the omission of the quantity of WDPS. We also analyzed the reasons behind it and provided potential solutions. The second part is extraction error, which demonstrates the extraction results of the FCN models have few connected, jagged, or perforated WDPS, which is beneficial for assessing fishery production, pattern changes, ecological value, and other applications of WDPS. The extracted WDPS by the FCN models are visually close to the ground truth, which is one of the most significant improvements over the traditional methods. The third part is special scenarios, including various shape types, intricate spatial configurations, and multiple pond conditions. WDPS with irregular shapes or juxtaposed with other land types increases the difficulty of extraction, but the FCN models still achieve P-scores above 0.95 in the first two scenarios, while WDPS in multiple pond conditions causes a sharp drop in the indicators of all the methods, which requires further improvement to solve it. We integrated the performances of the methods to provide recommendations for their use. This study offers valuable insights for enhancing deep learning methods and leveraging extraction results in practical applications. Full article

Accurate forecasting of multivariate traffic flow poses formidable challenges, primarily due to the ever-evolving spatio-temporal dynamics and intricate spatial heterogeneity, where the heterogeneity signifies that the correlations among locations are not just related to distance. However, few of the existing models are designed to fully and effectively integrate the above-mentioned features. To address these complexities head-on, this paper introduces a novel solution in the form of Dynamic Pattern-aware Spatio-Temporal Convolutional Networks (DPSTCN). Temporally, the model introduces a novel temporal module, containing a temporal convolutional network (TCN) enriched with an enhanced pattern-aware self-attention mechanism, adept at capturing temporal patterns, including local/global dependencies, dynamics, and periodicity. Spatially, the model constructs static and dynamic pattern-aware convolutions, leveraging geographical and area-functional information to effectively capture intricate spatial patterns, including dynamics and heterogeneity. Evaluations across four distinct traffic benchmark datasets consistently demonstrate the state-of-the-art capacity of our model compared to the existing eleven approaches, especially great improvements in RMSE (Root Mean Squared Error) value. Full article

An algorithmic model of a neural network with channel attention and spatial attention (CASANet) is proposed to estimate the value of atmospheric coherence length, which in turn provides a quantitative description of atmospheric turbulence intensity. By processing the acquired spot image data, the channel attention and spatial attention mechanisms are utilized, and the convolutional neural network learns the interdependence between the channel and space of the feature image and adaptively recalibrates the feature response in terms of the channel to increase the contribution of the foreground spot and suppress the background features. Based on the experimental data, an analysis of the CASANet model subject to turbulence intensity perturbations, fluctuations in outgoing power, and fluctuations in beam quality at the outlet is carried out. Comparison of the results of the convolutional neural network with those of the inverse method and the differential image motion method shows that the convolutional neural network is optimal in three evaluation indexes, namely, the mean deviation, the root-mean-square error, and the correlation coefficient, which are 2.74, 3.35, and 0.94, respectively. The convolutional neural network exhibits high accuracy under moderate and weak turbulence, and the estimation values under strong turbulence conditions are still mostly within the 95% confidence interval. The above results fully demonstrate that the proposed convolutional neural network method can effectively estimate the atmospheric coherence length, which provides technical support for the inversion of atmospheric turbulence intensity based on images. Full article

This paper proposes a short-term electric load forecasting method combining convolutional neural networks and gated recurrent units with an attention mechanism. By integrating CNNs and GRUs, the method can fully leverage the strengths of CNNs in feature extraction and the advantages of GRUs in sequence modeling, enabling the model to comprehend signal data more comprehensively and effectively extract features from sequential data. The introduction of the attention mechanism allows the traditional model to dynamically focus on important parts of the input data while ignoring the unimportant parts. This capability enables the model to utilize input information more efficiently, thereby enhancing model performance. This paper applies the proposed model to a dataset comprising regional electric load and meteorological data for experimentation. The results show that compared with other common models, the proposed model effectively reduces the mean absolute error and root mean square error (121.51 and 263.43, respectively) and accurately predicts the short-term change in regional power load. Full article

This work proposes and describes a deep learning-based approach utilizing recurrent neural networks (RNNs) for beam pattern synthesis considering uniform linear arrays. In this particular case, the deep neural network (DNN) learns from previously optimized radiation patterns as inputs and generates complex excitations as output. Beam patterns are optimized using a genetic algorithm during the training phase in order to reduce sidelobes and achieve high directivity. Idealized and test beam patterns are employed as inputs for the DNN, demonstrating their effectiveness in scenarios with high prediction complexity and closely spaced elements. Additionally, a comparative analysis is conducted among the three DNN architectures. Numerical experiments reveal improvements in performance when using the long short-term memory network (LSTM) compared to fully connected and convolutional neural networks. Full article

This study introduces an advanced deep-learning framework for the real-time detection of pipeline leaks in smart city infrastructure. The methodology transforms acoustic emission (AE) signals from the time domain into scalogram images using continuous wavelet transform (CWT) to enhance leak-related features. A Gaussian filter minimizes background noise and clarifies these features further. The core of the framework combines convolutional neural networks (CNNs) with long short-term memory (LSTM), ensuring a comprehensive examination of both spatial and temporal features of AE signals. A genetic algorithm (GA) optimizes the neural network by isolating the most important features for leak detection. The final classification stage uses a fully connected neural network to categorize pipeline health conditions as either ‘leak’ or ‘non-leak’. Experimental validation on real-world pipeline data demonstrated the framework’s efficacy, achieving accuracy rates of 99.69%. This approach significantly advances smart city capabilities in pipeline monitoring and maintenance, offering a durable and scalable solution for proactive infrastructure management. Full article

Reconfigurable processor-based acceleration of deep convolutional neural network (DCNN) algorithms has emerged as a widely adopted technique, with particular attention on sparse neural network acceleration as an active research area. However, many computing devices that claim high computational power still struggle to execute neural network algorithms with optimal efficiency, low latency, and minimal power consumption. Consequently, there remains significant potential for further exploration into improving the efficiency, latency, and power consumption of neural network accelerators across diverse computational scenarios. This paper investigates three key techniques for hardware acceleration of sparse neural networks. The main contributions are as follows: (1) Most neural network inference tasks are typically executed on general-purpose computing devices, which often fail to deliver high energy efficiency and are not well-suited for accelerating sparse convolutional models. In this work, we propose a specialized computational circuit for the convolutional operations of sparse neural networks. This circuit is designed to detect and eliminate the computational effort associated with zero values in the sparse convolutional kernels, thereby enhancing energy efficiency. (2) The data access patterns in convolutional neural networks introduce significant pressure on the high-latency off-chip memory access process. Due to issues such as data discontinuity, the data reading unit often fails to fully exploit the available bandwidth during off-chip read and write operations. In this paper, we analyze bandwidth utilization in the context of convolutional accelerator data handling and propose a strategy to improve off-chip access efficiency. Specifically, we leverage a compiler optimization plugin developed for Vitis HLS, which automatically identifies and optimizes on-chip bandwidth utilization. (3) In coefficient-based accelerators, the synchronous operation of individual computational units can significantly hinder efficiency. Previous approaches have achieved asynchronous convolution by designing separate memory units for each computational unit; however, this method consumes a substantial amount of on-chip memory resources. To address this issue, we propose a shared feature map cache design for asynchronous convolution in the accelerators presented in this paper. This design resolves address access conflicts when multiple computational units concurrently access a set of caches by utilizing a hash-based address indexing algorithm. Moreover, the shared cache architecture reduces data redundancy and conserves on-chip resources. Using the optimized accelerator, we successfully executed ResNet50 inference on an Intel Arria 10 1150GX FPGA, achieving a throughput of 497 GOPS, or an equivalent computational power of 1579 GOPS, with a power consumption of only 22 watts. Full article