Convolution Technique Research Articles

LDDP-Net: A Lightweight Neural Network with Dual Decoding Paths for Defect Segmentation of LED Chips

Chip defect detection is a crucial aspect of the semiconductor production industry, given its significant impact on chip performance. This paper proposes a lightweight neural network with dual decoding paths for LED chip segmentation, named LDDP-Net. Within the LDDP-Net framework, the receptive field of the MobileNetv3 backbone is modified to mitigate information loss. In addition, dual decoding paths consisting of a coarse decoding path and a fine-grained decoding path in parallel are developed. Specifically, the former employs a straightforward upsampling approach, emphasizing macro information. The latter is more detail-oriented, using multiple pooling and convolution techniques to focus on fine-grained information after deconvolution. Moreover, the integration of intermediate-layer features into the upsampling operation enhances boundary segmentation. Experimental results demonstrate that LDDP-Net achieves an mIoU (mean Intersection over Union) of 90.29% on the chip dataset, with parameter numbers and FLOPs (Floating Point Operations) of 2.98 M and 2.24 G, respectively. Comparative analyses with advanced methods reveal varying degrees of improvement, affirming the effectiveness of the proposed method.

Improved region-based active contour segmentation through divergence and convolution techniques

<p>In this paper, we present a novel approach to improve the robustness of region-based active contour models for image segmentation, particularly in the presence of noise. Traditional active contour methods often struggle with noise sensitivity and intensity variations within the image. To overcome these limitations, we propose an enhanced segmentation model that integrates the average convolution with entropy-based mean gray level values. Our method leverages the local statistical information by introducing a local similarity factor and local region relative entropy to build a robust energy functional. This energy functional balances the intensity differences between neighboring pixels and regions within the local window, while reducing the impact of noise. By incorporating convolution and entropy into the energy formulation, our model distinguishes between the interior and exterior regions of an image more effectively, thus leading to more accurate segmentation results. We demonstrate the numerical implementation of the proposed model, along with its convexity properties, to ensure stability and reliability. The experimental results show that our method significantly improves the segmentation performance, even in challenging scenarios with varying noise levels. This advancement has the potential to improve image analyses in fields such as medical imaging, object detection, and texture classification.</p>

Cardiac arrhythmia classification of imbalanced data using convolutional autoencoder and LSTM techniques

Cardiac arrhythmia classification of imbalanced data using convolutional autoencoder and LSTM techniques

Facial Recognition from Video Using Enhanced Social Collie Optimization-Based Deep Convolutional Neural Network Technique

Facial Recognition from Video Using Enhanced Social Collie Optimization-Based Deep Convolutional Neural Network Technique

The effectiveness of fire detection with convolutional neural networks

The article presents problems related to the detection of fire phenomena using convolutional neural network techniques. The main issue described in the article focuses on determining the precision of flame detection depending on lighting conditions and the selection of CNN architecture. The types of neural networks tested are primarily SSD architectures, which, with their speed of operation and energy consumption, are the most common in mobile applications. The study shows which of the neural network architectures used have the highest average precision in detecting the fire phenomenon. The selection of networks under testing was analyzed in terms of the speed of the algorithm and its precision. Four pre-trained neural network models were used during the testing of two training bases. The complexity of each model directly affected the training time of the model, which oscillated between 2-8 [h], and the precision achieved.

Graph Neural Networks on SPD Manifolds for Motor Imagery Classification: A Perspective From the Time-Frequency Analysis.

The motor imagery (MI) classification has been a prominent research topic in brain-computer interfaces (BCIs) based on electroencephalography (EEG). Over the past few decades, the performance of MI-EEG classifiers has seen gradual enhancement. In this study, we amplify the geometric deep-learning-based MI-EEG classifiers from the perspective of time-frequency analysis, introducing a new architecture called Graph-CSPNet. We refer to this category of classifiers as Geometric Classifiers, highlighting their foundation in differential geometry stemming from EEG spatial covariance matrices. Graph-CSPNet utilizes novel manifold-valued graph convolutional techniques to capture the EEG features in the time-frequency domain, offering heightened flexibility in signal segmentation for capturing localized fluctuations. To evaluate the effectiveness of Graph-CSPNet, we employ five commonly used publicly available MI-EEG datasets, achieving near-optimal classification accuracies in nine out of 11 scenarios. The Python repository can be found at https://github.com/GeometricBCI/Tensor-CSPNet-and-Graph-CSPNet.

DeepNeXt: a lightweight polyp segmentation algorithm based on multi-scale attention.

Fast and accurate automatic segmentation of polyps in colonoscopy plays a crucial role in the early diagnosis and treatment of colon cancer. However, the current polyp segmentation algorithms based on deep neural networks suffer from the problems of larger models and lower segmentation accuracy. Meanwhile, achieving accurate segmentation of polyps is to improve the diagnostic efficiency of doctors, and this need motivates us to develop a set of lightweight models so that it can be easily embedded in clinical devices to meet the requirements of practical applications. This study aims to provide effective technical support for the rapid and precise segmentation of polyps in clinical applications. DeepNeXt, an innovative polyp segmentation model grounded in multi-scale attention mechanisms. DeepNeXt incorporates a multi-stage, lightweight convolutional encoder module, leveraging several lightweight convolutional layers for efficient and accurate feature extraction. Furthermore, it features a novel multi-stage feature fusion structure designed to circumvent the potential loss of feature information during the encoding phase. Additionally, the model employs a multi-scale attentional feature encoding module that harnesses multi-branch deep strip convolution techniques to extract multi-dimensional information from the feature maps post-encoding, thereby enhancing the neural network's capability to extract diverse feature information. Experimental validation on the Kvasir Segmentation Dataset (Kvasir-SEG dataset) and Colorectal Cancer-Clinic Datasetbase (CVC-ClinicDB datasets) demonstrates that DeepNeXt outperforms mainstream networks such as U-net, U-net++, TransUnet, SwinUnet, and TGANet in terms of parameters and floating-point operations (FLOPs). DeepNeXt achieved a FLOPs metric of only 3.04 G and a parameters (Params) metric of just 1.51 M, while delivering exceptional performance in segmentation. On the Kvasir-SEG dataset, it reached a mean intersection over union (mIOU) of 83.91, and on the CVC-ClinicDB dataset, 87.37. Additionally, the Dice and Recall metrics also showed superior results, highlighting that DeepNeXt strikes an optimal balance between computational efficiency, model compactness, and segmentation accuracy. In conclusion, we have proposed the DeepNeXt network, a novel lightweight multi-scale attention segmentation network tailored for computationally limited medical devices, which provides strong support for accurate and efficient polyp segmentation in clinical applications.

Design of Multimedia Teaching Software Based on Distributed Network Architecture

With the leap of network technology and the vigorous development of online teaching, many universities are actively adopting online means to optimize teaching and course management. This paper focuses on building an efficient and comprehensive auxiliary teaching platform that integrates functions such as student learning monitoring, course management, online testing, and teacher–student interaction, aiming to improve the quality of education. Designs based on distributed B/S architecture and web technology to ensure efficient resource allocation and expansion, meet the needs of large-scale concurrent learning, and achieve cross-platform access to enhance user experience. The platform features personalized learning support, enhanced interactive collaboration, and the construction of a multimedia teaching database through data analysis. The innovation lies in using neural network technology to create an intelligent question answering module, utilizing cosine similarity to automatically group teaching resources, and using graph convolution and variational autoencoder techniques to construct a student performance monitoring model. Experimental verification shows that the platform runs stably, effectively reduces the blocking rate, and significantly improves students’ academic performance, pass rate, and learning interest. This design not only successfully completed the teaching task, but also provided valuable experience for the application of online-assisted teaching systems in subject education.

A simplified approach for simulating pollutant transport in small rivers with dead zones using convolution

Abstract In the paper an alternative method to solve the one-dimensional advective-diffusive equation describing the pollutants transport in river with dead zones is presented. Because very often transport in a small river can be treated as a 1D issue, then instead of numerical solution of the advection-diffusion equation an equivalent approach based on the convolution technique can be used. Consequently, for a given impulse response function the numerical calculations are required to compute a convolution only. The impulse response function is obtained as an analytical solution of the linear advection-diffusion equation for the Dirac delta function imposed as the boundary condition at the upstream end. Therefore, it represents the Gauss distribution and consequently, this approach is unreliable when the dead zones occur. To reproduce an asymmetric distribution of concentration along the channel axis an approximation of analytical impulse response function using the asymmetric Gumbel distribution is proposed. This approach valid for solution of the transport equation with constant coefficients is extended for piecewise constant coefficients. Convolution approach does not produce any numerical dissipation and dispersion errors typically generated by the methods based on the finite difference technique. Validation of the method using the results of field measurements confirmed its effectiveness.

Lightweight UAV Small Target Detection and Perception Based on Improved YOLOv8-E

Traditional unmanned aerial vehicle (UAV) detection methods struggle with multi-scale variations during flight, complex backgrounds, and low accuracy, whereas existing deep learning detection methods have high accuracy but high dependence on equipment, making it difficult to detect small UAV targets efficiently. To address the above challenges, this paper proposes an improved lightweight high-precision model, YOLOv8-E (Enhanced YOLOv8), for the fast and accurate detection and identification of small UAVs in complex environments. First, a Sobel filter is introduced to enhance the C2f module to form the C2f-ESCFFM (Edge-Sensitive Cross-Stage Feature Fusion Module) module, which achieves higher computational efficiency and feature representation capacity while preserving detection accuracy as much as possible by fusing the SobelConv branch for edge extraction and the convolution branch to extract spatial information. Second, the neck network is based on the HSFPN (High-level Screening-feature Pyramid Network) architecture, and the CAA (Context Anchor Attention) mechanism is introduced to enhance the semantic parsing of low-level features to form a new CAHS-FPN (Context-Augmented Hierarchical Scale Feature Pyramid Network) network, enabling the fusion of deep and shallow features. This improves the feature representation capability of the model, allowing it to detect targets of different sizes efficiently. Finally, the optimized detail-enhanced convolution (DEConv) technique is introduced into the head network, forming the LSCOD (Lightweight Shared Convolutional Object Detector Head) module, enhancing the generalization ability of the model by integrating a priori information and adopting the strategy of shared convolution. This ensures that the model enhances its localization and classification performance without increasing parameters or computational costs, thus effectively improving the detection performance of small UAV targets. The experimental results show that compared with the baseline model, the YOLOv8-E model achieved (mean average precision at IoU = 0.5) an mAP@0.5 improvement of 6.3%, reaching 98.4%, whereas the model parameter scale was reduced by more than 50%. Overall, YOLOv8-E significantly reduces the demand for computational resources while ensuring high-precision detection.

Application of Convolutional Neural Network (CNN) and different other techniques for the restoration of degraded folk artworks: a comparative performance analysis

Application of Convolutional Neural Network (CNN) and different other techniques for the restoration of degraded folk artworks: a comparative performance analysis

Graph Neural Networks with maximal independent set-based pooling: Mitigating over-smoothing and over-squashing

Graph Neural Networks (GNNs) have significantly advanced graph-level prediction tasks by utilizing efficient convolution and pooling techniques. However, traditional pooling methods in GNNs often fail to preserve key properties, leading to challenges such as graph disconnection, low decimation ratios, and substantial data loss. In this paper, we introduce three novel pooling methods based on Maximal Independent Sets (MIS) to address these issues. Additionally, we provide a theoretical and empirical study on the impact of these pooling methods on over-smoothing and over-squashing phenomena. Our experimental results not only confirm the effectiveness of using maximal independent sets to define pooling operations but also demonstrate their crucial role in mitigating over-smoothing and over-squashing.

Asymmetric Large Kernel Distillation Network for efficient single image super-resolution.

Recently, significant advancements have been made in the field of efficient single-image super-resolution, primarily driven by the innovative concept of information distillation. This method adeptly leverages multi-level features to facilitate high-resolution image reconstruction, allowing for enhanced detail and clarity. However, many existing approaches predominantly emphasize the enhancement of distilled features, often overlooking the critical aspect of improving the feature extraction capabilities of the distillation module itself. In this paper, we address this limitation by introducing an asymmetric large-kernel convolution design. By increasing the size of the convolution kernel, we expand the receptive field, which enables the model to more effectively capture long-range dependencies among image pixels. This enhancement significantly improves the model's perceptual ability, leading to more accurate reconstructions. To maintain a manageable level of model complexity, we adopt a lightweight architecture that employs asymmetric convolution techniques. Building on this foundation, we propose the Lightweight Asymmetric Large Kernel Distillation Network (ALKDNet). Comprehensive experiments conducted on five widely recognized benchmark datasets-Set5, Set14, BSD100, Urban100, and Manga109-indicate that ALKDNet not only preserves efficiency but also demonstrates performance enhancements relative to existing super-resolution methods. The average PSNR and SSIM values show improvements of 0.10 dB and 0.0013, respectively, thereby achieving state-of-the art performance.

Refined feature enhancement network for object detection

Convolutional neural networks-based object detection techniques have achieved positive performances. However, due to the limitations of local receptive field, some existing object detection methods cannot effectively capture global information in feature extraction phases, and thus lead to unsatisfactory detection performance. Moreover, the feature information extracted by the backbone network may be redundant. To alleviate these problems, in this paper we propose a refined feature enhancement network (RFENet) for object detection. Specifically, we first propose a feature enhancement module (FEM) to capture more global and local information from feature maps with certain long-range dependencies. We further propose a multi-branch dilated attention mechanism (MDAM) to refine the extracted features in a weighted form, which can select more important spatial and channel information and broaden the receptive field of the network. Finally, we validate RFENet on MS-COCO2017, PASCAL VOC2012, and PASCAL VOC07+12 datasets, respectively. Compared to the baseline network, our RFENet improves by 2.4 AP on MS-COCO2017 dataset, 3.4 mAP on PASCAL VOC2012 dataset, and 2.7 mAP on PASCAL VOC07+12 dataset. Extensive experiments show that our RFENet can perform competitively on different datasets. The code is available at https://github.com/object9detection/RFENet.

An improved face attributes editing method based on DDIM

The main advantage of DDIM is that it guarantees the quality of the generated images while increasing the efficiency of the generation by modifying the sampling strategy in the diffusion process. DiffusionRig, which addresses the problem of maintaining identity consistency by learning the person-specific facial prior in a tiny personalized dataset, is a successful representation of the DDIM strategy. Based on DiffusionRig, in this article, we propose an improved face attributes editing method based on DDIM to improve naturalness and accuracy of editing results in complex face attribute editing tasks and the generalization ability. Our method combines DDIM and DECA together and use two-stage training strategy. To reduce DiffusionRig’s limitations in handling face attribute editing tasks that require nonlinear understanding and fine-tuning, our method also introduces a channel attention mechanism and a depth-separable convolution technique in the training model. In the first stage we trained our model on the open FFHQ dataset, which consists of 30,000 high-resolution face image. In the second stage, the model was refined by using a tiny personalized dataset. Face attribute editing experiments,comparative experiment with DiffusionRig, and a series of ablation experiments have been performed. Then, the validity of our improved method is verified by qualitative and quantitative analysis.

DSCIMABNet: A novel multi-head attention depthwise separable CNN model for skin cancer detection

Skin cancer is a common type of cancer worldwide. Early diagnosis of skin cancer can reduce the risk of death by increasing treatment success. However, it is challenging for dermatologists or specialists because the symptoms are vague in the early stages and cannot be noticed by the naked eye. This study examines digital diagnostic techniques supported by artificial intelligence, focusing on early skin cancer detection and two methods have been proposed. In the first method, DSCIMABNet deep learning architecture was developed by combining multi-head attention and depthwise separable convolution techniques. This model provides flexibility in learning the dataset's local features, abstract concepts, and long-term relationships. The DSCIMABNet model and modern deep learning models trained on ImageNet are proposed to be combined with the ensemble learning method in the second method. This approach provides a comprehensive feature extraction process that will increase the performance of the classification process with ensemble learning. The proposed approaches are trained and evaluated on the ISIC 2018 dataset with image enhancement applied in preprocessing. In the experimental results, DSCIMABNet achieved 84.28% accuracy, while the proposed hybrid method achieved 99.40% accuracy. Moreover, on the Mendeley dataset (CNN for Melanoma Detection Data), DSCIMABNet achieved 92.58% accuracy, while the hybrid method achieved 99.37% accuracy. This study may significantly contribute to developing new and effective methods for the early diagnosis and treatment of skin cancer.

Automated image acquisition and analysis of graphene and hexagonal boron nitride from pristine to highly defective and amorphous structures

Defect-engineered and even amorphous two-dimensional (2D) materials have recently gained interest due to properties that differ from their pristine counterparts. Since these properties are highly sensitive to the exact atomic structure, it is crucial to be able to characterize them at atomic resolution over large areas. This is only possible when the imaging process is automated to reduce the time spent on manual imaging, which at the same time reduces the observer bias in selecting the imaged areas. Since the necessary datasets include at least hundreds if not thousands of images, the analysis process similarly needs to be automated. Here, we introduce disorder into graphene and monolayer hexagonal boron nitride (hBN) using low-energy argon ion irradiation, and characterize the resulting disordered structures using automated scanning transmission electron microscopy annular dark field imaging combined with convolutional neural network-based analysis techniques. We show that disorder manifests in these materials in a markedly different way, where graphene accommodates vacancy-type defects by transforming hexagonal carbon rings into other polygonal shapes, whereas in hBN the disorder is observed simply as vacant lattice sites with very little rearrangement of the remaining atoms. Correspondingly, in the case of graphene, the highest introduced disorder leads to an amorphous membrane, whereas in hBN, the highly defective lattice contains a large number of vacancies and small pores with no indication of amorphisation. Overall, our study demonstrates that combining automated imaging and image analysis is a powerful way to characterize the structure of disordered and amorphous 2D materials, while also illustrating some of the remaining shortcomings with this methodology.

ECG signal reconstruction from PPG using a hybrid attention-based deep learning network

Electrocardiography (ECG) and photoplethysmography (PPG) are non-invasive methods used to quantify signals originating from the cardiovascular system. Although there is a strong correlation between the cycles of the two measures, the correlation between the waveforms has received little attention in research. Measuring the PPG is significantly simpler and more convenient compared to the ECG measure. Recent research has demonstrated that PPG signals can be utilized to rebuild the ECG signals, suggesting that healthcare professionals might acquire a comprehensive comprehension of patients' cardiovascular well-being by only measuring PPG. With the advancement of artificial intelligence, the deep learning model has made significant progress in reconstructing ECG signals from PPG signals. However, the quality of the reconstructed ECG signal is still in need of more enhancement. In order to enhance the quality of the reconstructed ECG signal, we propose an innovative hybrid attention-based bidirectional recurrent neural network (Bi-RNN) that incorporates a dilated convolutional neural network technique for the purpose of reconstructing ECG signals from PPG signals. This addresses issues that occur when standard dilated CNN (DCNN) models neglect the link between circumstances and dispersion of gradients. The suggested approach maximizes the utilization of the DCNN and Bi-RNN unit design to provide fusion features. In order to ensure the model's resilience to deformation, it is imperative to initially extract spatial properties utilizing convolutional neural networks (CNNs). Once we have obtained geographic data, we employ BiLSTM to extract temporal features. The proposed model is so-called hybrid attention-based CNN and BiLSTM (HA-CNN-BiLSTM). BiLSTM mitigates the issues of gradient vanishing and exploding while maintaining accuracy. To showcase the advantages of the suggested model, we conduct a comparison between HA-CNN-BiLSTM and each separate state-of-the-art method. The simulation results indicated that the proposed method yielded superior root mean square error (RMSE) when reconstructing the ECG signal across different optimizers. The results also demonstrated that the proposed method results in the lowest RMSE with SGDM optimizer which is 0.031.

Pressure and Temperature Prediction of Oil Pipeline Networks Based on a Mechanism-Data Hybrid Driven Method

To ensure the operational safety of oil transportation stations, it is crucial to predict the impact of pressure and temperature before crude oil enters the pipeline network. Accurate predictions enable the assessment of the pipeline’s load-bearing capacity and the prevention of potential safety incidents. Most existing studies primarily focus on describing and modeling the mechanisms of the oil flow process. However, monitoring data can be skewed by factors such as instrument aging and pipeline friction, leading to inaccurate predictions when relying solely on mechanistic or data-driven approaches. To address these limitations, this paper proposes a Temporal-Spatial Three-stream Temporal Convolutional Network (TS-TTCN) model that integrates mechanistic knowledge with data-driven methods. Building upon Temporal Convolutional Networks (TCN), the TS-TTCN model synthesizes mechanistic insights into the oil transport process to establish a hybrid driving mechanism. In the temporal dimension, it incorporates real-time operating parameters and applies temporal convolution techniques to capture the time-series characteristics of the oil transportation pipeline network. In the spatial dimension, it constructs a directed topological map based on the pipeline network’s node structure to characterize spatial features. Data analysis and experimental results show that the Three-stream Temporal Convolutional Network (TTCN) model, which uses a Tanh activation function, achieves an error rate below 5%. By analyzing and validating real-time data from the Dongying oil transportation station, the proposed hybrid model proves to be more stable, reliable, and accurate under varying operating conditions.

Enhancing Generalizability in Biomedical Entity Recognition: Self-Attention PCA-CLS Model.

One of the primary tasks in the early stages of data mining involves the identification of entities from biomedical corpora. Traditional approaches relying on robust feature engineering face challenges when learning from available (un-)annotated data using data-driven models like deep learning-based architectures. Despite leveraging large corpora and advanced deep learning models, domain generalization remains an issue. Attention mechanisms are effective in capturing longer sentence dependencies and extracting semantic and syntactic information from limited annotated datasets. To address out-of-vocabulary challenges in biomedical text, the PCA-CLS (Position and Contextual Attention with CNN-LSTM-Softmax) model combines global self-attention and character-level convolutional neural network techniques. The model's performance is evaluated on eight distinct biomedical domain datasets encompassing entities such as genes, drugs, diseases, and species. The PCA-CLS model outperforms several state-of-the-art models, achieving notable F1-scores, including 88.19% on BC2GM, 85.44% on JNLPBA, 90.80% on BC5CDR-chemical, 87.07% on BC5CDR-disease, 89.18% on BC4CHEMD, 88.81% on NCBI, and 91.59% on the s800 dataset.