Multi-scale Network Research Articles

Multi-scale differential network for landslide extraction from remote sensing images with different scenarios

ABSTRACT Landslides are major geological hazards globally, causing significant economic losses each year. Accurate landslide detection is essential for disaster prevention, risk assessment, and timely emergency response. Current extraction methods struggle to distinguish landslides from their surroundings and precisely define their boundaries. To address these challenges, we introduce the Multi-Scale Difference Enhancement Network (MSDENet), a framework for landslide extraction through time-based change detection. MSDENet incorporates three core components: the Difference Guided Attention Module (DGAM) for enhanced focus on landslide-specific changes, the Multi-Scale Feature Fusion Module (MSFFM) for improved boundary delineation, and the Multi-Scale Sensory Module (MSSM) to boost generalization by integrating multi-scale features. We validate MSDENet’s effectiveness on the Global Very-High-Resolution Landslide Mapping (GVLM) dataset, covering 17 diverse landslide events, and further assess its applicability on high-resolution Nepal and Wenchuan datasets. MSDENet outperforms six contemporary frameworks, achieving IoU improvements of 1.42% and 1.08% for the Kaikoura and Tbilisi datasets and demonstrating gains of 3.97% and 4.79% for the Nepal and Wenchuan datasets, confirming its effectiveness in varied conditions.

Accurate detection of arbitrary ship directions using SAR based on RTMDet

ABSTRACT Ship detection, an important branch of Synthetic Aperture Radar (SAR) images interpretation, is crucial for various maritime reconnaissance and surveillance but remains challenging due to background interference, dense ship alignments and high aspect ratios. To address these challenges, this letter proposes an arbitrary direction detection method based on Real-Time Models for object Detection (RTMDet), which is one of the fastest and most accurate single-stage detection detectors. Firstly, an attention selection module is designed to replace traditional full convolution method to achieve more parameter reduction; Secondly, in order to extract more key ship feature information and accelerate the network extraction, we construct a lightweight multi-scale feature pyramid network; Thirdly, a novel loss function is introduced to enhance the detection of rotated bounding box by addressing the problem of bounding box discontinuities and lack of scale invariance. The proposed method achieves state-of-the-art detection accuracy on two the high-resolution SAR image datasets for arbitrary ship detection including HRSID (85.6% AP50) and RSDD (90.8% AP50), while demonstrating significant accuracy improvements in the complex inshore scenes.

A multi-scale temporal-frequency fusion network based on MLP for long-term time series forecasting

A multi-scale temporal-frequency fusion network based on MLP for long-term time series forecasting

Fault diagnosis method of rolling bearing under variable operating conditions based on MFCCNN

A rolling bearing fault diagnosis method based on multi-scale feature fusion and cross-level connection convolutional neural network (MFCCNN) is proposed to address the issues of low accuracy and low generalization performance caused by the complex and variable actual working conditions of rolling bearings. Firstly, a Markov transition field (MTF) is used to convert a one-dimensional signal into a two-dimensional image with temporal correlation. Next, an improved channel attention mechanism is integrated into the multi-scale network to assign different weights to each channel information and extract initial features. Then, a position attention mechanism is created and incorporated into the cross-level connection to enhance the feature expression ability for deep feature extraction. The MFCCNN faults diagnosis model is developed based on the above module. Finally, the MTF images is input into the proposed model for training, and the Softmax classifier is utilized to achieve fault classification. To verify the effectiveness and superiority of the MFCCNN method, the CWRU and MFS rolling bearing dataset are selected for experimental verification. The results demonstrates that the MFCCNN model has higher diagnostic accuracy, stronger noise immunity, and generalization performance in varying operating conditions compared to other fault diagnosis methods.

Multi-Scale Network Analysis of Community-Based Senior Centers: Exploring the Intersection of Spatial Embeddedness and Accessibility in Nanjing, China

As critical infrastructure of age-friendly cities, senior centers are designed to be embedded in communities and facilitate service accessibility for older adults. However, their underutilization is widely documented, suggesting a need to reassess their effectiveness. Existing studies often analyze the issue focusing on socio-demographic factors, overlooking the spatial contexts in which senior centers are embedded and their impacts on older people’s access. This study aims to address the research gap by investigating the spatial embeddedness of senior centers using Space Syntax methods and examining its influence on older people’s access patterns. Using a geo-behavioral dataset collected in Nanjing, China, we find that about 70% of the senior centers in the research area are embedded in highly localized settings with limited connections to global street networks, which significantly restricts the access of older people from wider spatial contexts. This spatial segregation may force senior centers to incur higher costs to attract users, thereby reducing the effectiveness of community-based services. This study introduces a novel spatial perspective to evaluate community-based services, highlighting the critical influence of the spatial context on service accessibility. The findings provide valuable empirical insights for policymakers and planners striving to create age-friendly cities and communities.

Analysis of Different Methods in Pedestrian Re-identification

This study focuses on the field of pedestrian re-identification (ReID), aiming to enhance the accuracy and efficiency of individual recognition through advanced deep learning models. The article first introduces two deep learning models: Omni-Scale Feature Learning for Person Re-Identification (OSNet) and Multi-Scale Interaction Network (MSINet). OSNet employs Depthwise Separable Convolutions (DSC) and Omni-Scale Residual Blocks to improve the ability to learn features across different scales. MSINet, on the other hand, utilizes neural architecture search technology to design a lightweight network architecture, enhancing feature discrimination and flexible utilization through Twins Contrastive Mechanism (TCM) and Multi-Scale Interaction (MSI). Additionally, a Spatial Alignment Module (SAM) is proposed to enhance the consistency of images under different viewpoints or conditions. The experimental section selects two widely used pedestrian re-identification datasets, Market1501 and MSMT17, for evaluation, and the results show that MSINet outperforms existing methods in terms of accuracy and stability. The article concludes by summarizing the advantages of OSNet and MSINet in multi-scale feature learning and points out their application limitations.

Construction of tennis pose estimation and action recognition model based on improved ST-GCN

With the rapid growth of computer vision and deep learning technologies, the application of pose estimation and action recognition in sports training has become increasingly widespread. Due to factors such as complex movements, fast speed, and limb occlusion, pose estimation and action recognition in tennis face significant challenges. Therefore, this study first introduces selective dropout and pyramid region of interest pooling layer strategies in fast region convolutional neural networks. Secondly, a pose estimation algorithm based on multi-scale fusion pose residual network 50 is designed, and finally a spatiotemporal graph convolutional network model is constructed by fusing channel attention module and multi-scale dilated convolution module. The data showed that the average detection accuracy of the improved attitude residual network 50 was 70.4%, and the accuracy of object detection for small, medium, and large objects was 57.4%, 69.3%, and 79.2%, respectively. The continuous action recognition accuracy and inter action fluency detection time of the improved spatiotemporal graph convolutional network were 93.8% and 19.2 ms, respectively. When the sample size was 1000, its memory usage was 1378 MB and the running time was 32.7 ms. Experiments have shown that the improved model achieves high accuracy and robustness in tennis action recognition tasks, especially in complex scenes and limb occlusion conditions, where the model shows significant advantages. This study aims to provide an efficient and accurate motion recognition technology for tennis posture analysis and intelligent training.

Polarization of road target detection under complex weather conditions

Polarization imaging technology can be applied to unveil the interaction between light and matter by harnessing the transverse vector wave attributes of light, thus to accentuate target characteristics amidst complex weather conditions. This technology has the potential to be widely used in road target detection. However, polarization detection is significantly affected by illumination and detection angles, as well as the considerable variation in the scale of road targets. The optimal polarization parameters should be adaptively adjusted to weather conditions, angles and target features, whereas most existing research employs handcrafted polarization parameters without considering actual complex detection requirements, which are unable to adaptively adjust the polarization feature enhancement methods. In this paper, we propose a road target detection algorithm based on an end-to-end adaptive polarization coding method, named YOLO-Polarization of Road Target Detection (YOLO-PRTD). To enhance the polarized features of targets under complex weather conditions, an Adaptive Polarization Coding Module (APCM) is designed. This module integrates channel-wise global self-attention and small kernel convolution to adaptively adjust the polarization enhancement method using dynamically extracted global and local polarization feature information. A multi-scale detection network is also designed to fully extract and fuse multi-scale feature information from receptive fields, channels, and spaces in different dimensions. Additionally, a dataset of Polarized Images of Road Targets in Complex Weather conditions (PIRT-CW) is proposed for training and evaluation. Experimental results on the PIRT-CW show that the YOLO-PRTD algorithm achieves a mAP0.5 of 89.83%, reducing the error rate by 15.54% compared to the baseline network YOLOX.

Integrating multi-scale attention network with deep image prior for single image restoration

Integrating multi-scale attention network with deep image prior for single image restoration

A multi-scale and multi-modal convolutional neural network for condition monitoring of transmission line

Abstract Monitoring the fatigue damage of transmission lines is crucial for stable power system operation. However, existing model-driven methods face challenges such as high computational complexity and reliance on expert knowledge, while data-driven methods require large amounts of abnormal state data. To address these issues, a multi-scale and multi-modal convolutional neural network (CNN) is proposed for real-time condition monitoring of transmission lines. Key steps include: firstly, empirical Fourier decomposition is used to decompose the original signals, extracting multi-scale state information at different frequency scales. Then, time-domain, frequency-domain, and time–frequency domain analyses are performed on the decomposed signals to capture multi-modal information. Based on this, a multi-modal fusion network is proposed based on a CNN to extract shallow and deep features, with a fully connected layer used for multi-modal feature fusion. Notably, the algorithm is implemented on a microprocessor for practical application. Experimental results show that the proposed model achieves a diagnostic accuracy of 93.06%, outperforming classical networks. It also surpasses models trained solely on time, frequency, or time–frequency features by 25.18%, 21.8%, and 19.3%, respectively.

Multi-scale Unet-based feature aggregation network for lightweight image deblurring

Multi-scale Unet-based feature aggregation network for lightweight image deblurring

EC-PFN: a multiscale woven fusion network for industrial product surface defect detection

In order to address challenges such as small target sizes, low contrast, significant intra-class variations, and indistinct inter-class differences in surface defect detection, this paper proposes the Enhanced Context-aware Parallel Fusion Network (EC-PFN). The network employs a Featur Weave Network architecture to enhance contextal awareess and parallel fusion capabilities. It utilizes a Feature Fusion Module (UniFusionLayer) for effective multiscale and multisemantic feature learning, offering new perspectives on feature fusion. Additionally, a Receptive Field Block (RFB) module is introduced to expand the receptive field, enhancing feature extraction in scenarios with low contrast and subtle defects. The Loss Ranking Module (LRM) is incorporated to optimize the target-oriented loss, improving performance by omitting low-confidence bounding boxes. Extensive experiments on a light guide plate defect dataset demonstrate that EC-PFN achieves a detection accuracy (mAP) of 98.9%, a detection speed of 92 FPS, and a computational cost of 14.5 GFLOPs, outperforming mainstream surface defect detection models.

An improved multi-scale feature extraction network for medical image segmentation.

The use of U-Net and its variations has led to significant advancements in medical image segmentation. However, the encoder-decoder structures of these models often lose spatial information during downsampling. Skip connections can help address this issue; however, they may also introduce excessive irrelevant background information. Additionally, medical images display significant scale variations and complex tissue structures, making it challenging for existing models to accurately separate tissues from the background. To address these issues, we developed the Res2Net-ConvFormer-Dilation-UNet (Res2-CD-UNet), a multi-scale feature extraction network for medical image segmentation. This study presents a novel U-shaped segmentation network that employs Res2Net as the backbone and incorporates a convolution-style transformer in the encoding stage to enhance global attention. Additionally, a novel channel feature fusion block (CFFB) has been introduced in the skip connection stage to minimize the effects of background noise. The proposed model was evaluated using publicly available datasets, Synapse and Seg.A.2023. Using the Synapse dataset, the average dice similarity coefficient (DSC) reached 83.92%, which was 1.96% higher than the suboptimal model, and the average Hausdorff distance (HD) was 14.51 mm. Among the eight organs evaluated, optimal results were achieved for four organs. Similarly, using the Seg.A.2023 dataset, the proposed model also achieved the best results with an average DSC of 93.27%. The results of this study indicate that the proposed model can more accurately segment regions of interest and better extract multi-scale features in medical images than existing deep-learning algorithms.

Multi-scale perceptual modulation network for low-dose computed tomography denoising.

Low-dose computed tomography (LDCT) reduces radiation exposure, but the introduced noise and artifacts impair its diagnostic accuracy. Convolutional neural networks (CNNs) are widely used for LDCT denoising, but they suffer from a limited receptive field. The use of a larger kernel size can enlarge the receptive field and boost model performance; however, the computational cost of the model greatly increases. We aimed to develop a LDCT denoising CNN with a large receptive field and lower computational complexity. We developed a multi-scale perceptual modulation network (MSPMnet) incorporating a powerful multi-head decomposable convolution (MHDC). To address the high computational complexity of large kernel convolutions, we developed a novel MHDC module that can capture multi-scale features and efficiently expand the receptive field. The MHDC module couples maximum-pooling with three depth-wise convolutions of varying kernel sizes via a channel splitting mechanism, where, unlike conventional CNNs, the two large two-dimensional kernels are each decomposed into a set of cascaded orthogonal one-dimensional kernels to remain lightweight. Further, departing from prior methodologies that apply a uniform kernel size throughout the network, we introduced a receptive field-ramp mechanism that adeptly transitions from local to relatively long-range dependency modeling as the network depth increases, thereby achieving superior performance. The proposed MSPMnet was evaluated on a Mayo Clinic data set with a conventional iterative algorithm, two CNN models, and two Transformer models used for comparison. Compared to the competing baseline methods, the MSPMnet exhibited better performance in both the visual and quantitative assessments. Visually, the MSPMnet preserved the structure, edges, and textures with excellent noise and artifact reduction, generating the denoised images closest to normal-dose computed tomography images. Quantitatively, the MSPMnet had the lowest root mean-square error (RMSE) (8.3094±1.9325) and the highest peak signal-to-noise ratio (PSNR) (33.8525±1.8213 dB), structural similarity index (SSIM) (0.9309±0.0272), and feature similarity index (FSIM) (0.9699±0.0113), demonstrating superior denoising performance. The proposed MSPMnet excelled at LDCT denoising, effectively removing noise and artifacts while preserving edges. Compared to the state-of-the-art CNNs and Transformers, the proposed MSPMnet exhibited superior denoising performance both quantitatively and qualitatively.

Surface roughness prediction based on fusion of dynamic-static data

The surface roughness in mechanical milling is a critical indicator of machining quality. Due to the variability in the geometric structure of the workpieces and the complexity of the milling mechanisms, traditional roughness prediction methods often fail to meet the requirements of actual machining scenarios. To enhance the accuracy of surface roughness prediction during the milling process and to ensure its applicability in real-world production, this paper introduces a novel model that integrates Multi-scale Convolutional Neural Network, Stacked Bidirectional Long Short-Term Memory, and Self-Attention mechanisms (MC-SBiL-SA), which integrates dynamic current signals with static process data for roughness prediction. This model involves extracting features from dynamic signal data in various ways, then feeding these features into a BiLSTM network for training. To enable the model to learn more effective data features, an attention mechanism is introduced after the BiLSTM to assign different weights to the data features, thereby amplifying the impact of important features. Finally, the dynamic signal and static process data are combined in a shallow neural network for regression prediction. Comparative results with various baseline models demonstrate that the MC-SBiL-SA model, which fuses dynamic and static data, exhibits superior generalizability and effectiveness in predicting surface roughness.

Single image super-resolution with lightweight multi-scale dilated attention network

Single image super-resolution with lightweight multi-scale dilated attention network

Geological remote sensing interpretation via a local-to-global sensitive feature fusion network

Interpreting surface geological elements (such as rocks, minerals, soils, and water bodies) is the main task of geological survey, which plays a crucial role in geological environment remote sensing (GERS). However, the characteristics of geological elements, including high variabilities, various morphology, complicated boundaries and imbalanced class distribution, make it still a challenge for deep learning methods to interpret GERS images. Considering the correlations of geological elements as the regionalized variables in geostatistics, the sensitive features of GERS interpretation mainly include three aspects: tonal, textural and structural characteristics within a singular-class elements, spatial and spectral correlations of adjacent elements, and their global tectonic or spatial distribution. Thus, to simulate the manual interpretation process of geologists from local to global and promote GERS interpretation performance, we propose a local-to-global multi-scale feature fusion network (LGMSFNet). A geological object context represents the intra-class semantic dependencies of pixel sets with the same class. And a local feature aggregation module models the channel and spatial association. Then discriminative features are integrated by a global feature fusion module. For the model optimization, we focus on hard examples during the training process to achieve the balanced optimization of various categories. Two research areas that include large-scale rocks, soils and water exposed on the surface are selected. Massive experiments demonstrate the superiority of the LGMSFNet in GERS interpretation.

Neutron-Image Super-Resolution based on Convolutional Neural Networks and Novel Parallel Transformer

Enhancing the resolution of neutron imaging can be achieved by upgrading the hardware components of the neutron imaging system, including employing more potent neutron source systems, utilizing high-resolution detectors, and implementing superior collimation techniques. However, such improvements are frequently constrained by challenges such as high cost and limited flexibility. In this paper, a flexible and efficient single-neutron-image super-resolution network is proposed for neutron image super-resolution. We propose a novel multi-scale feature extraction network, which can extract features deeply by fusing features of different types and levels. In order to learn more context information, a novel parallel Transformer is proposed to overcome the long dependency problem of sequences. A high-frequency information enhancement network is proposed to enhance the high-frequency information of neutron image in order to address the blur problem. Quantitative and qualitative analyses of benchmark datasets and real neutron images show that the proposed network performs better on Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM) than bicubic interpolation and other methods, and reconstructs high-resolution neutron images that are more consistent with human visual perception, highlighting its application potential in neutron images quality enhancement.

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.

DTTCNet: Time-to-Collision Estimation With Autonomous Emergency Braking Using Multi-Scale Transformer Network

DTTCNet: Time-to-Collision Estimation With Autonomous Emergency Braking Using Multi-Scale Transformer Network