An improved model based on you only look once version 8(YOLOv8)is proposed to solve the problem of low detection accuracy due to the diversity of object sizes in optical remote sensing images.Firstly,the feature pyram...An improved model based on you only look once version 8(YOLOv8)is proposed to solve the problem of low detection accuracy due to the diversity of object sizes in optical remote sensing images.Firstly,the feature pyramid network(FPN)structure of the original YOLOv8 mode is replaced by the generalized-FPN(GFPN)structure in GiraffeDet to realize the"cross-layer"and"cross-scale"adaptive feature fusion,to enrich the semantic information and spatial information on the feature map to improve the target detection ability of the model.Secondly,a pyramid-pool module of multi atrous spatial pyramid pooling(MASPP)is designed by using the idea of atrous convolution and feature pyramid structure to extract multi-scale features,so as to improve the processing ability of the model for multi-scale objects.The experimental results show that the detection accuracy of the improved YOLOv8 model on DIOR dataset is 92%and mean average precision(mAP)is 87.9%,respectively 3.5%and 1.7%higher than those of the original model.It is proved the detection and classification ability of the proposed model on multi-dimensional optical remote sensing target has been improved.展开更多
Multi-label image classification is a challenging task due to the diverse sizes and complex backgrounds of objects in images.Obtaining class-specific precise representations at different scales is a key aspect of feat...Multi-label image classification is a challenging task due to the diverse sizes and complex backgrounds of objects in images.Obtaining class-specific precise representations at different scales is a key aspect of feature representation.However,existing methods often rely on the single-scale deep feature,neglecting shallow and deeper layer features,which poses challenges when predicting objects of varying scales within the same image.Although some studies have explored multi-scale features,they rarely address the flow of information between scales or efficiently obtain class-specific precise representations for features at different scales.To address these issues,we propose a two-stage,three-branch Transformer-based framework.The first stage incorporates multi-scale image feature extraction and hierarchical scale attention.This design enables the model to consider objects at various scales while enhancing the flow of information across different feature scales,improving the model’s generalization to diverse object scales.The second stage includes a global feature enhancement module and a region selection module.The global feature enhancement module strengthens interconnections between different image regions,mitigating the issue of incomplete represen-tations,while the region selection module models the cross-modal relationships between image features and labels.Together,these components enable the efficient acquisition of class-specific precise feature representations.Extensive experiments on public datasets,including COCO2014,VOC2007,and VOC2012,demonstrate the effectiveness of our proposed method.Our approach achieves consistent performance gains of 0.3%,0.4%,and 0.2%over state-of-the-art methods on the three datasets,respectively.These results validate the reliability and superiority of our approach for multi-label image classification.展开更多
Cracks represent a significant hazard to pavement integrity,making their efficient and automated extraction essential for effective road health monitoring and maintenance.In response to this challenge,we propose a cra...Cracks represent a significant hazard to pavement integrity,making their efficient and automated extraction essential for effective road health monitoring and maintenance.In response to this challenge,we propose a crack automatic extraction network model that integrates multi⁃scale image features,thereby enhancing the model’s capability to capture crack characteristics and adaptation to complex scenarios.This model is based on the ResUNet architecture,makes modification to the convolutional layer of the model,proposes to construct multiple branches utilizing different convolution kernel sizes,and adds a atrous spatial pyramid pooling module within the intermediate layers.In this paper,comparative experiments on the performance of the basic model,ablation experiments,comparative experiments before and after data augmentation,and generalization verification experiments are conducted.Comparative experimental results indicate that the improved model exhibits superior detail processing capability at crack edges.The overall performance of the model,as measured by the F1⁃score,reaches 71.03%,reflecting a 2.1%improvement over the conventional ResUNet.展开更多
We propose a hierarchical multi-scale attention mechanism-based model in response to the low accuracy and inefficient manual classification of existing oceanic biological image classification methods. Firstly, the hie...We propose a hierarchical multi-scale attention mechanism-based model in response to the low accuracy and inefficient manual classification of existing oceanic biological image classification methods. Firstly, the hierarchical efficient multi-scale attention(H-EMA) module is designed for lightweight feature extraction, achieving outstanding performance at a relatively low cost. Secondly, an improved EfficientNetV2 block is used to integrate information from different scales better and enhance inter-layer message passing. Furthermore, introducing the convolutional block attention module(CBAM) enhances the model's perception of critical features, optimizing its generalization ability. Lastly, Focal Loss is introduced to adjust the weights of complex samples to address the issue of imbalanced categories in the dataset, further improving the model's performance. The model achieved 96.11% accuracy on the intertidal marine organism dataset of Nanji Islands and 84.78% accuracy on the CIFAR-100 dataset, demonstrating its strong generalization ability to meet the demands of oceanic biological image classification.展开更多
To improve image quality under low illumination conditions,a novel low-light image enhancement method is proposed in this paper based on multi-illumination estimation and multi-scale fusion(MIMS).Firstly,the illuminat...To improve image quality under low illumination conditions,a novel low-light image enhancement method is proposed in this paper based on multi-illumination estimation and multi-scale fusion(MIMS).Firstly,the illumination is processed by contrast-limited adaptive histogram equalization(CLAHE),adaptive complementary gamma function(ACG),and adaptive detail preserving S-curve(ADPS),respectively,to obtain three components.Then,the fusion-relevant features,exposure,and color contrast are selected as the weight maps.Subsequently,these components and weight maps are fused through multi-scale to generate enhanced illumination.Finally,the enhanced images are obtained by multiplying the enhanced illumination and reflectance.Compared with existing approaches,this proposed method achieves an average increase of 0.81%and 2.89%in the structural similarity index measurement(SSIM)and peak signal-to-noise ratio(PSNR),and a decrease of 6.17%and 32.61%in the natural image quality evaluator(NIQE)and gradient magnitude similarity deviation(GMSD),respectively.展开更多
Aiming at the problem of low detection accuracy due to the different scale sizes of apple leaf disease spots and their similarity to the background,this paper proposes a multi-scale lightweight network(MSL-Net).Firstl...Aiming at the problem of low detection accuracy due to the different scale sizes of apple leaf disease spots and their similarity to the background,this paper proposes a multi-scale lightweight network(MSL-Net).Firstly,a multiplexed aggregated feature extraction network is proposed using residual bottleneck block(RES-Bottleneck)and middle partial-convolution(MP-Conv)to capture multi-scale spatial features and enhance focus on disease features for better differentiation between disease targets and background information.Secondly,a lightweight feature fusion network is designed using scale-fuse concatenation(SF-Cat)and triple-scale sequence feature fusion(TSSF)module to merge multi-scale feature maps comprehensively.Depthwise convolution(DWConv)and GhostNet lighten the network,while the cross stage partial bottleneck with 3 convolutions ghost-normalization attention module(C3-GN)reduces missed detections by suppressing irrelevant background information.Finally,soft non-maximum suppression(Soft-NMS)is used in the post-processing stage to improve the problem of misdetection of dense disease sites.The results show that the MSL-Net improves mean average precision at intersection over union of 0.5(mAP@0.5)by 2.0%over the baseline you only look once version 5s(YOLOv5s)and reduces parameters by 44%,reducing computation by 27%,outperforming other state-of-the-art(SOTA)models overall.This method also shows excellent performance compared to the latest research.展开更多
Objectives:Schizophrenia is a profoundly stigmatized mental health condition,characterized by misconceptions that affect societal attitudes,policy development,and the lived experiences of individuals with the conditio...Objectives:Schizophrenia is a profoundly stigmatized mental health condition,characterized by misconceptions that affect societal attitudes,policy development,and the lived experiences of individuals with the condition.This study aimed to develop and validate a multidimensional scale for assessing societal stigma towards schizophrenia,while exploring how demographic factors influence such attitudes.Methods:Drawing on an extensive literature review and consultations,the study identified five domains of stigma:Workplace Capability,Intimate Relationships,Autonomy,Risk Perception,and Recovery.Using a two-phase methodology,a preliminary 38-itemscale was administered to 729 participants from the general Spanish population,refining the measure through descriptive and exploratory factor analysis.Subsequently,a revised 34-item scale was validated through confirmatory factor analysis with an independent sample of 417 participants.Results:The final model showed good fit(RMSEA=0.056,CFI=0.938,TLI=0.933)and strong internal consistency(α=0.73–0.86).Findings revealed that stigma was most pronounced in the domain of Autonomy(Mean=2.83,SD=0.91),reflecting pervasive doubts about individuals’ability to live independently and achieve meaningful integration into society.Stigma varied significantly across demographic variables,with higher levels reported among men,older individuals,married participants,and those outside health professions(p<0.01).Conversely,healthcare professionals,younger individuals,and those familiar with someone with schizophrenia generally reported less stigma(p<0.01).Conclusion:This study developed and validated a robust multidimensional scale for assessing societal stigma toward schizophrenia.The five-factor model—Workplace Capability,Intimate Relationships,Autonomy,Risk Perception,and Recovery—was empirically supported.Autonomy and Recovery emerged as themost stigmatized domains across the Spanish general population.The scale demonstrated strong psychometric properties and effectively captured stigma patterns linked to key sociodemographic variables.展开更多
Large interfacial strains in particles are crucial for promoting bonding in cold spraying(CS),initiated either by adiabatic shear instability(ASI)due to softening prevailing over strain hardening or by hydrostatic pla...Large interfacial strains in particles are crucial for promoting bonding in cold spraying(CS),initiated either by adiabatic shear instability(ASI)due to softening prevailing over strain hardening or by hydrostatic plasticity,which is claimed to promote bonding even without ASI.A thorough microstructural analysis is vital to fully understand the bonding mechanisms at play during microparticle impacts and throughout the CS process.In this study,the HEA CoCrFeMnNi,known for its relatively high strain hardening and resistance to softening,was selected to investigate the microstructure characteristics and bonding mech-anisms in CS.This study used characterization techniques covering a range of length scales,including electron channeling contrast imaging(ECCI),electron backscatter diffraction(EBSD),and high-resolution transmission microscopy(HR-TEM),to explore the microstructure characteristics of bonding and overall structure development of CoCrFeMnNi microparticles after impact in CS.HR-TEM lamellae were prepared using focused ion beam milling.Additionally,the effects of deformation field variables on microstructure development were determined through finite element modeling(FEM)of microparticle impacts.The ECCI,EBSD,and HR-TEM analyses revealed an interplay between dislocation-driven processes and twinning,leading to the development of four distinct deformation microstructures.Significant grain refinement occurs at the interface through continuous dynamic recrystallization(CDRX)due to high strain and temperature rise from adiabatic deformation,signs of softening,and ASI.Near the interface,a necklace-like structure of refined grains forms around grain boundaries,along with elongated grains,resulting from the coexistence of dynamic recovery and discontinuous dynamic recrystallization(DDRX)due to lower temperature rise and strain.Towards the particle or substrate interior,concurrent twinning and dislocation-mediated mechanisms refine the structure,forming straight,curved,and intersected twins.At the top of the particles,only deformed grains with a low dislocation density are observed.Our results showed that DRX induces microstructure softening in highly strained interface areas,facilitating atomic bonding in CoCrFeMnNi.HR-TEM investigation confirms the formation of atomic bonds between particles and substrate,with a gradual change in crystal lattice orientation from the particle to the substrate and the occurrence of some misfit dislocations and vacancies at the interface.Finally,the findings of this research suggest that softening and ASI,even in materials resistant to softening,are required to establish bonding in CS.展开更多
This paper focuses on the problem of traffic flow forecasting,with the aim of forecasting future traffic conditions based on historical traffic data.This problem is typically tackled by utilizing spatio-temporal graph...This paper focuses on the problem of traffic flow forecasting,with the aim of forecasting future traffic conditions based on historical traffic data.This problem is typically tackled by utilizing spatio-temporal graph neural networks to model the intricate spatio-temporal correlations among traffic data.Although these methods have achieved performance improvements,they often suffer from the following limitations:These methods face challenges in modeling high-order correlations between nodes.These methods overlook the interactions between nodes at different scales.To tackle these issues,in this paper,we propose a novel model named multi-scale dynamic hypergraph convolutional network(MSDHGCN)for traffic flow forecasting.Our MSDHGCN can effectively model the dynamic higher-order relationships between nodes at multiple time scales,thereby enhancing the capability for traffic forecasting.Experiments on two real-world datasets demonstrate the effectiveness of the proposed method.展开更多
As optimization problems continue to grow in complexity,the need for effective metaheuristic algorithms becomes increasingly evident.However,the challenge lies in identifying the right parameters and strategies for th...As optimization problems continue to grow in complexity,the need for effective metaheuristic algorithms becomes increasingly evident.However,the challenge lies in identifying the right parameters and strategies for these algorithms.In this paper,we introduce the adaptive multi-strategy Rabbit Algorithm(RA).RA is inspired by the social interactions of rabbits,incorporating elements such as exploration,exploitation,and adaptation to address optimization challenges.It employs three distinct subgroups,comprising male,female,and child rabbits,to execute a multi-strategy search.Key parameters,including distance factor,balance factor,and learning factor,strike a balance between precision and computational efficiency.We offer practical recommendations for fine-tuning five essential RA parameters,making them versatile and independent.RA is capable of autonomously selecting adaptive parameter settings and mutation strategies,enabling it to successfully tackle a range of 17 CEC05 benchmark functions with dimensions scaling up to 5000.The results underscore RA’s superior performance in large-scale optimization tasks,surpassing other state-of-the-art metaheuristics in convergence speed,computational precision,and scalability.Finally,RA has demonstrated its proficiency in solving complicated optimization problems in real-world engineering by completing 10 problems in CEC2020.展开更多
With the continuous evolution of urban surface types,the impact of the urban heat island effect on the human population has intensified.Investigating the factors influencing urban thermal environments is crucial for p...With the continuous evolution of urban surface types,the impact of the urban heat island effect on the human population has intensified.Investigating the factors influencing urban thermal environments is crucial for providing theoretical support to urban planning and decision-making.In this study,Shenyang was selected to comprehensively analyse multiple factors,including topography,human activity,vegetation and landscape.Moreover,we used the random forest algorithm to explore nonlinear factors influencing land surface temperature(LST)over four years in the study area.The results revealed that from 2005 to 2020,the total areas with sub-high and high-temperature zones in northern Shenyang steadily increased.The area ratio of these zones increased from 20.18% in 2005 to 24.86% in 2020.Additionally,significant and strong correlations were observed between LST and variables such as the enhanced vegetation index(EVI),normalised difference vegetation index(NDVI),population density,proportion of cropland and proportion of impervious land.In 2010,proportion of impervious land exhibited the strongest correlation with LST at the 5 km scale,reaching 0.852(p<0.01).The 4 km grid scale was identified as the optimal grid size for this study,while the 2 km grid performed the worst.In 2020,NDVI emerged as the most significant factor influencing LST.These findings provide valuable guidance for improving urban planning and developing sustainable strategies.展开更多
Lost circulation critically jeopardizes drilling safety and efficiency,and remains an unresolved challenge in oil and gas engineering.In this paper,by utilizing the self-developed dynamic plugging apparatus and synthe...Lost circulation critically jeopardizes drilling safety and efficiency,and remains an unresolved challenge in oil and gas engineering.In this paper,by utilizing the self-developed dynamic plugging apparatus and synthetic cores containing large-scale fractures,experimental research on the circulation plugging of different materials was conducted.Based on the D90 rule and fracture mechanical aperture model,we analyze the location of plugging layer under dynamic plugging mechanism.By setting different parameters of fracture width and injection pressure,the laws of cyclic plugging time,pressure bearing capacity and plugging layers formation were investigated.The results show that the comprehensive analysis of particle size and fracture aperture provides an accurate judgment of the entrance-plugging phenomenon.The bridging of solid materials in the leakage channel is a gradual process,and the formation of a stable plug requires 2–3 plug-leakage cycles.The first and second cyclic plugging time was positively correlated with the fracture width.Different scales of fractures were successfully plugged with the bearing pressure greater than 6 MPa,but there were significant differences in the composition of the plugging layer.The experimental results can effectively prove that the utilized plugging agent is effective and provides an effective reference for dynamic plugging operation.展开更多
The application of ecosystem services(ES)theories in land consolidation is a confusing issue that has long plagued scholars and government officials.As the upgraded version of traditional land consolidation,comprehens...The application of ecosystem services(ES)theories in land consolidation is a confusing issue that has long plagued scholars and government officials.As the upgraded version of traditional land consolidation,comprehensive land consolidation(CLC)emphasizes ecological benefits,but it does not achieve the expected effect during the pilot phase.This study first proposed a theoretical analysis framework based on ES knowledge to answer the three key questions of why,where,and how to implement CLC better.Taking mountainous counties as the study area,we found that ES trade-offs/synergies,bundles,and drivers were significantly affected by scale effects.ES knowledge can play a crucial role in designing multi-scale CLC strategies regarding the objective,zoning,intensity,and mode.Specifically,mitigating the significant trade-offs between recreational opportunities,food production,and other ES is the top priority of CLC.Land consolidation zoning based on the ES bundles analysis is more rational and can provide the scientific premise for designing locally adapted CLC measures.Land consolidation can be classified into high-intensity direct intervention and low-intensity indirect intervention modes,based on the major drivers of ES.These findings help narrow the gap between ES and CLC practices.展开更多
As the use of deepfake facial videos proliferate,the associated threats to social security and integrity cannot be overstated.Effective methods for detecting forged facial videos are thus urgently needed.While many de...As the use of deepfake facial videos proliferate,the associated threats to social security and integrity cannot be overstated.Effective methods for detecting forged facial videos are thus urgently needed.While many deep learning-based facial forgery detection approaches show promise,they often fail to delve deeply into the complex relationships between image features and forgery indicators,limiting their effectiveness to specific forgery techniques.To address this challenge,we propose a dual-branch collaborative deepfake detection network.The network processes video frame images as input,where a specialized noise extraction module initially extracts the noise feature maps.Subsequently,the original facial images and corresponding noise maps are directed into two parallel feature extraction branches to concurrently learn texture and noise forgery clues.An attention mechanism is employed between the two branches to facilitate mutual guidance and enhancement of texture and noise features across four different scales.This dual-modal feature integration enhances sensitivity to forgery artifacts and boosts generalization ability across various forgery techniques.Features from both branches are then effectively combined and processed through a multi-layer perception layer to distinguish between real and forged video.Experimental results on benchmark deepfake detection datasets demonstrate that our approach outperforms existing state-of-the-art methods in terms of detection performance,accuracy,and generalization ability.展开更多
Objective To report the development,validation,and findings of the Multi-dimensional Attention Rating Scale(MARS),a self-report tool crafted to evaluate six-dimension attention levels.Methods The MARS was developed ba...Objective To report the development,validation,and findings of the Multi-dimensional Attention Rating Scale(MARS),a self-report tool crafted to evaluate six-dimension attention levels.Methods The MARS was developed based on Classical Test Theory(CTT).Totally 202 highly educated healthy adult participants were recruited for reliability and validity tests.Reliability was measured using Cronbach's alpha and test-retest reliability.Structural validity was explored using principal component analysis.Criterion validity was analyzed by correlating MARS scores with the Toronto Hospital Alertness Test(THAT),the Attentional Control Scale(ACS),and the Attention Network Test(ANT).Results The MARS comprises 12 items spanning six distinct dimensions of attention:focused attention,sustained attention,shifting attention,selective attention,divided attention,and response inhibition.As assessed by six experts,the content validation index(CVI)was 0.95,the Cronbach's alpha for the MARS was 0.78,and the test-retest reliability was 0.81.Four factors were identified(cumulative variance contribution rate 68.79%).The total score of MARS was correlated positively with THAT(r=0.60,P<0.01)and ACS(r=0.78,P<0.01)and negatively with ANT's reaction time for alerting(r=−0.31,P=0.049).Conclusion The MARS can reliably and validly assess six-dimension attention levels in real-world settings and is expected to be a new tool for assessing multi-dimensional attention impairments in different mental disorders.展开更多
Urban sprawl is a critical challenge in the urban development trajectory of developing countries,necessitating precise measurement,trend projection,and strategic management to achieve sustainable urban growth.This stu...Urban sprawl is a critical challenge in the urban development trajectory of developing countries,necessitating precise measurement,trend projection,and strategic management to achieve sustainable urban growth.This study focuses on the Yangtze River Economic Belt(YREB)as a case region and introduces a comprehensive evaluation framework that incorporates multidimensional factors and addresses the scale effects of urban sprawl.We emphasize the value of a systematic geographical approach by quantifying urban sprawl through simulated scenarios and analyzing its driving factors.We constructed an innovative urban sprawl index(USI)to assess the degree of sprawl within the YREB.This assessment integrates two geographic models with an artificial neural network algorithm,enabling simulation of urban sprawl trends under two future scenarios for 2035.Additionally,two analytical methods were employed to identify the key driving mechanisms of urban sprawl in the region.Findings indicate a strong correlation between urban scale and the extent of urban sprawl:larger urban areas exhibit more pronounced sprawl,with agglomeration and morphological transformations identified as primary contributors to urban sprawl.The study further reveals an intricate association between urban sprawl and the compactness of urban internal structures.While both development scenarios offer distinct advantages,the Coordinated Development Scenario is projected to foster a more balanced urban expansion.The robustness of the evaluation framework was enhanced through simulation and an in-depth analysis of internal mechanisms,bolstering confidence in its applicability.We advocate for the adoption and continued refinement of this framework as a tool for promoting balanced urban growth.The strategic recommendations provided herein are vital for mitigating multi-scale urban sprawl,advancing economic development,and improving residents’quality of life across cities in the YREB.展开更多
Electrocatalysis has been extensively explored for the storage and conversion of renewable electric power.Understanding the physisorption and chemisorption processes at electrified solid–liquid interfaces(ESLIs)is cr...Electrocatalysis has been extensively explored for the storage and conversion of renewable electric power.Understanding the physisorption and chemisorption processes at electrified solid–liquid interfaces(ESLIs)is crucial for revealing the typical surface restructuring and catalyst dissolution during electrocatalysis.Although advanced in situ tools and theoretical models have been proposed[1,2],identifying the nature of the active sites with atomic-scale spatial resolution remains a challenge,especially at ESLIs.In a recent work published in Nature,Zhang et al.[3]reported a groundbreaking atomic-resolution imaging of the structural dynamics of Cu nanowire catalysts in ESLIs for electrochemical CO_(2)reduction(ECR).展开更多
To address the issues of frequent identity switches(IDs)and degraded identification accuracy in multi object tracking(MOT)under complex occlusion scenarios,this study proposes an occlusion-robust tracking framework ba...To address the issues of frequent identity switches(IDs)and degraded identification accuracy in multi object tracking(MOT)under complex occlusion scenarios,this study proposes an occlusion-robust tracking framework based on face-pedestrian joint feature modeling.By constructing a joint tracking model centered on“intra-class independent tracking+cross-category dynamic binding”,designing a multi-modal matching metric with spatio-temporal and appearance constraints,and innovatively introducing a cross-category feature mutual verification mechanism and a dual matching strategy,this work effectively resolves performance degradation in traditional single-category tracking methods caused by short-term occlusion,cross-camera tracking,and crowded environments.Experiments on the Chokepoint_Face_Pedestrian_Track test set demonstrate that in complex scenes,the proposed method improves Face-Pedestrian Matching F1 area under the curve(F1 AUC)by approximately 4 to 43 percentage points compared to several traditional methods.The joint tracking model achieves overall performance metrics of IDF1:85.1825%and MOTA:86.5956%,representing improvements of 0.91 and 0.06 percentage points,respectively,over the baseline model.Ablation studies confirm the effectiveness of key modules such as the Intersection over Area(IoA)/Intersection over Union(IoU)joint metric and dynamic threshold adjustment,validating the significant role of the cross-category identity matching mechanism in enhancing tracking stability.Our_model shows a 16.7%frame per second(FPS)drop vs.fairness of detection and re-identification in multiple object tracking(FairMOT),with its cross-category binding module adding aboute 10%overhead,yet maintains near-real-time performance for essential face-pedestrian tracking at small resolutions.展开更多
High-throughput transcriptomics has evolved from bulk RNA-seq to single-cell and spatial profiling,yet its clinical translation still depends on effective integration across diverse omics and data modalities.Emerging ...High-throughput transcriptomics has evolved from bulk RNA-seq to single-cell and spatial profiling,yet its clinical translation still depends on effective integration across diverse omics and data modalities.Emerging foundation models and multimodal learning frameworks are enabling scalable and transferable representations of cellular states,while advances in interpretability and real-world data integration are bridging the gap between discovery and clinical application.This paper outlines a concise roadmap for AI-driven,transcriptome-centered multi-omics integration in precision medicine(Figure 1).展开更多
基金supported by the National Natural Science Foundation of China(No.62241109)the Tianjin Science and Technology Commissioner Project(No.20YDTPJC01110)。
文摘An improved model based on you only look once version 8(YOLOv8)is proposed to solve the problem of low detection accuracy due to the diversity of object sizes in optical remote sensing images.Firstly,the feature pyramid network(FPN)structure of the original YOLOv8 mode is replaced by the generalized-FPN(GFPN)structure in GiraffeDet to realize the"cross-layer"and"cross-scale"adaptive feature fusion,to enrich the semantic information and spatial information on the feature map to improve the target detection ability of the model.Secondly,a pyramid-pool module of multi atrous spatial pyramid pooling(MASPP)is designed by using the idea of atrous convolution and feature pyramid structure to extract multi-scale features,so as to improve the processing ability of the model for multi-scale objects.The experimental results show that the detection accuracy of the improved YOLOv8 model on DIOR dataset is 92%and mean average precision(mAP)is 87.9%,respectively 3.5%and 1.7%higher than those of the original model.It is proved the detection and classification ability of the proposed model on multi-dimensional optical remote sensing target has been improved.
基金supported by the National Natural Science Foundation of China(62302167,62477013)Natural Science Foundation of Shanghai(No.24ZR1456100)+1 种基金Science and Technology Commission of Shanghai Municipality(No.24DZ2305900)the Shanghai Municipal Special Fund for Promoting High-Quality Development of Industries(2211106).
文摘Multi-label image classification is a challenging task due to the diverse sizes and complex backgrounds of objects in images.Obtaining class-specific precise representations at different scales is a key aspect of feature representation.However,existing methods often rely on the single-scale deep feature,neglecting shallow and deeper layer features,which poses challenges when predicting objects of varying scales within the same image.Although some studies have explored multi-scale features,they rarely address the flow of information between scales or efficiently obtain class-specific precise representations for features at different scales.To address these issues,we propose a two-stage,three-branch Transformer-based framework.The first stage incorporates multi-scale image feature extraction and hierarchical scale attention.This design enables the model to consider objects at various scales while enhancing the flow of information across different feature scales,improving the model’s generalization to diverse object scales.The second stage includes a global feature enhancement module and a region selection module.The global feature enhancement module strengthens interconnections between different image regions,mitigating the issue of incomplete represen-tations,while the region selection module models the cross-modal relationships between image features and labels.Together,these components enable the efficient acquisition of class-specific precise feature representations.Extensive experiments on public datasets,including COCO2014,VOC2007,and VOC2012,demonstrate the effectiveness of our proposed method.Our approach achieves consistent performance gains of 0.3%,0.4%,and 0.2%over state-of-the-art methods on the three datasets,respectively.These results validate the reliability and superiority of our approach for multi-label image classification.
基金supported in part by the National Natural Science Foundation of China(No.42401166)the Open Fund of Key Laboratory of Polar Environment Monitoring and Public Governance,Ministry of Education(No.202405)the Key Research and Development Program of Hebei Province(No.23375405D).
文摘Cracks represent a significant hazard to pavement integrity,making their efficient and automated extraction essential for effective road health monitoring and maintenance.In response to this challenge,we propose a crack automatic extraction network model that integrates multi⁃scale image features,thereby enhancing the model’s capability to capture crack characteristics and adaptation to complex scenarios.This model is based on the ResUNet architecture,makes modification to the convolutional layer of the model,proposes to construct multiple branches utilizing different convolution kernel sizes,and adds a atrous spatial pyramid pooling module within the intermediate layers.In this paper,comparative experiments on the performance of the basic model,ablation experiments,comparative experiments before and after data augmentation,and generalization verification experiments are conducted.Comparative experimental results indicate that the improved model exhibits superior detail processing capability at crack edges.The overall performance of the model,as measured by the F1⁃score,reaches 71.03%,reflecting a 2.1%improvement over the conventional ResUNet.
基金supported by the National Natural Science Foundation of China (Nos.61806107 and 61702135)。
文摘We propose a hierarchical multi-scale attention mechanism-based model in response to the low accuracy and inefficient manual classification of existing oceanic biological image classification methods. Firstly, the hierarchical efficient multi-scale attention(H-EMA) module is designed for lightweight feature extraction, achieving outstanding performance at a relatively low cost. Secondly, an improved EfficientNetV2 block is used to integrate information from different scales better and enhance inter-layer message passing. Furthermore, introducing the convolutional block attention module(CBAM) enhances the model's perception of critical features, optimizing its generalization ability. Lastly, Focal Loss is introduced to adjust the weights of complex samples to address the issue of imbalanced categories in the dataset, further improving the model's performance. The model achieved 96.11% accuracy on the intertidal marine organism dataset of Nanji Islands and 84.78% accuracy on the CIFAR-100 dataset, demonstrating its strong generalization ability to meet the demands of oceanic biological image classification.
基金supported by the National Key R&D Program of China(No.2022YFB3205101)NSAF(No.U2230116)。
文摘To improve image quality under low illumination conditions,a novel low-light image enhancement method is proposed in this paper based on multi-illumination estimation and multi-scale fusion(MIMS).Firstly,the illumination is processed by contrast-limited adaptive histogram equalization(CLAHE),adaptive complementary gamma function(ACG),and adaptive detail preserving S-curve(ADPS),respectively,to obtain three components.Then,the fusion-relevant features,exposure,and color contrast are selected as the weight maps.Subsequently,these components and weight maps are fused through multi-scale to generate enhanced illumination.Finally,the enhanced images are obtained by multiplying the enhanced illumination and reflectance.Compared with existing approaches,this proposed method achieves an average increase of 0.81%and 2.89%in the structural similarity index measurement(SSIM)and peak signal-to-noise ratio(PSNR),and a decrease of 6.17%and 32.61%in the natural image quality evaluator(NIQE)and gradient magnitude similarity deviation(GMSD),respectively.
文摘Aiming at the problem of low detection accuracy due to the different scale sizes of apple leaf disease spots and their similarity to the background,this paper proposes a multi-scale lightweight network(MSL-Net).Firstly,a multiplexed aggregated feature extraction network is proposed using residual bottleneck block(RES-Bottleneck)and middle partial-convolution(MP-Conv)to capture multi-scale spatial features and enhance focus on disease features for better differentiation between disease targets and background information.Secondly,a lightweight feature fusion network is designed using scale-fuse concatenation(SF-Cat)and triple-scale sequence feature fusion(TSSF)module to merge multi-scale feature maps comprehensively.Depthwise convolution(DWConv)and GhostNet lighten the network,while the cross stage partial bottleneck with 3 convolutions ghost-normalization attention module(C3-GN)reduces missed detections by suppressing irrelevant background information.Finally,soft non-maximum suppression(Soft-NMS)is used in the post-processing stage to improve the problem of misdetection of dense disease sites.The results show that the MSL-Net improves mean average precision at intersection over union of 0.5(mAP@0.5)by 2.0%over the baseline you only look once version 5s(YOLOv5s)and reduces parameters by 44%,reducing computation by 27%,outperforming other state-of-the-art(SOTA)models overall.This method also shows excellent performance compared to the latest research.
文摘Objectives:Schizophrenia is a profoundly stigmatized mental health condition,characterized by misconceptions that affect societal attitudes,policy development,and the lived experiences of individuals with the condition.This study aimed to develop and validate a multidimensional scale for assessing societal stigma towards schizophrenia,while exploring how demographic factors influence such attitudes.Methods:Drawing on an extensive literature review and consultations,the study identified five domains of stigma:Workplace Capability,Intimate Relationships,Autonomy,Risk Perception,and Recovery.Using a two-phase methodology,a preliminary 38-itemscale was administered to 729 participants from the general Spanish population,refining the measure through descriptive and exploratory factor analysis.Subsequently,a revised 34-item scale was validated through confirmatory factor analysis with an independent sample of 417 participants.Results:The final model showed good fit(RMSEA=0.056,CFI=0.938,TLI=0.933)and strong internal consistency(α=0.73–0.86).Findings revealed that stigma was most pronounced in the domain of Autonomy(Mean=2.83,SD=0.91),reflecting pervasive doubts about individuals’ability to live independently and achieve meaningful integration into society.Stigma varied significantly across demographic variables,with higher levels reported among men,older individuals,married participants,and those outside health professions(p<0.01).Conversely,healthcare professionals,younger individuals,and those familiar with someone with schizophrenia generally reported less stigma(p<0.01).Conclusion:This study developed and validated a robust multidimensional scale for assessing societal stigma toward schizophrenia.The five-factor model—Workplace Capability,Intimate Relationships,Autonomy,Risk Perception,and Recovery—was empirically supported.Autonomy and Recovery emerged as themost stigmatized domains across the Spanish general population.The scale demonstrated strong psychometric properties and effectively captured stigma patterns linked to key sociodemographic variables.
文摘Large interfacial strains in particles are crucial for promoting bonding in cold spraying(CS),initiated either by adiabatic shear instability(ASI)due to softening prevailing over strain hardening or by hydrostatic plasticity,which is claimed to promote bonding even without ASI.A thorough microstructural analysis is vital to fully understand the bonding mechanisms at play during microparticle impacts and throughout the CS process.In this study,the HEA CoCrFeMnNi,known for its relatively high strain hardening and resistance to softening,was selected to investigate the microstructure characteristics and bonding mech-anisms in CS.This study used characterization techniques covering a range of length scales,including electron channeling contrast imaging(ECCI),electron backscatter diffraction(EBSD),and high-resolution transmission microscopy(HR-TEM),to explore the microstructure characteristics of bonding and overall structure development of CoCrFeMnNi microparticles after impact in CS.HR-TEM lamellae were prepared using focused ion beam milling.Additionally,the effects of deformation field variables on microstructure development were determined through finite element modeling(FEM)of microparticle impacts.The ECCI,EBSD,and HR-TEM analyses revealed an interplay between dislocation-driven processes and twinning,leading to the development of four distinct deformation microstructures.Significant grain refinement occurs at the interface through continuous dynamic recrystallization(CDRX)due to high strain and temperature rise from adiabatic deformation,signs of softening,and ASI.Near the interface,a necklace-like structure of refined grains forms around grain boundaries,along with elongated grains,resulting from the coexistence of dynamic recovery and discontinuous dynamic recrystallization(DDRX)due to lower temperature rise and strain.Towards the particle or substrate interior,concurrent twinning and dislocation-mediated mechanisms refine the structure,forming straight,curved,and intersected twins.At the top of the particles,only deformed grains with a low dislocation density are observed.Our results showed that DRX induces microstructure softening in highly strained interface areas,facilitating atomic bonding in CoCrFeMnNi.HR-TEM investigation confirms the formation of atomic bonds between particles and substrate,with a gradual change in crystal lattice orientation from the particle to the substrate and the occurrence of some misfit dislocations and vacancies at the interface.Finally,the findings of this research suggest that softening and ASI,even in materials resistant to softening,are required to establish bonding in CS.
基金the National Key Research and Development Program of China(No.2021ZD0112400)。
文摘This paper focuses on the problem of traffic flow forecasting,with the aim of forecasting future traffic conditions based on historical traffic data.This problem is typically tackled by utilizing spatio-temporal graph neural networks to model the intricate spatio-temporal correlations among traffic data.Although these methods have achieved performance improvements,they often suffer from the following limitations:These methods face challenges in modeling high-order correlations between nodes.These methods overlook the interactions between nodes at different scales.To tackle these issues,in this paper,we propose a novel model named multi-scale dynamic hypergraph convolutional network(MSDHGCN)for traffic flow forecasting.Our MSDHGCN can effectively model the dynamic higher-order relationships between nodes at multiple time scales,thereby enhancing the capability for traffic forecasting.Experiments on two real-world datasets demonstrate the effectiveness of the proposed method.
文摘As optimization problems continue to grow in complexity,the need for effective metaheuristic algorithms becomes increasingly evident.However,the challenge lies in identifying the right parameters and strategies for these algorithms.In this paper,we introduce the adaptive multi-strategy Rabbit Algorithm(RA).RA is inspired by the social interactions of rabbits,incorporating elements such as exploration,exploitation,and adaptation to address optimization challenges.It employs three distinct subgroups,comprising male,female,and child rabbits,to execute a multi-strategy search.Key parameters,including distance factor,balance factor,and learning factor,strike a balance between precision and computational efficiency.We offer practical recommendations for fine-tuning five essential RA parameters,making them versatile and independent.RA is capable of autonomously selecting adaptive parameter settings and mutation strategies,enabling it to successfully tackle a range of 17 CEC05 benchmark functions with dimensions scaling up to 5000.The results underscore RA’s superior performance in large-scale optimization tasks,surpassing other state-of-the-art metaheuristics in convergence speed,computational precision,and scalability.Finally,RA has demonstrated its proficiency in solving complicated optimization problems in real-world engineering by completing 10 problems in CEC2020.
基金National Natural Science Foundation of China,No.42204031。
文摘With the continuous evolution of urban surface types,the impact of the urban heat island effect on the human population has intensified.Investigating the factors influencing urban thermal environments is crucial for providing theoretical support to urban planning and decision-making.In this study,Shenyang was selected to comprehensively analyse multiple factors,including topography,human activity,vegetation and landscape.Moreover,we used the random forest algorithm to explore nonlinear factors influencing land surface temperature(LST)over four years in the study area.The results revealed that from 2005 to 2020,the total areas with sub-high and high-temperature zones in northern Shenyang steadily increased.The area ratio of these zones increased from 20.18% in 2005 to 24.86% in 2020.Additionally,significant and strong correlations were observed between LST and variables such as the enhanced vegetation index(EVI),normalised difference vegetation index(NDVI),population density,proportion of cropland and proportion of impervious land.In 2010,proportion of impervious land exhibited the strongest correlation with LST at the 5 km scale,reaching 0.852(p<0.01).The 4 km grid scale was identified as the optimal grid size for this study,while the 2 km grid performed the worst.In 2020,NDVI emerged as the most significant factor influencing LST.These findings provide valuable guidance for improving urban planning and developing sustainable strategies.
基金financially supported by National Natural Science Foundation of China(No.52422402)。
文摘Lost circulation critically jeopardizes drilling safety and efficiency,and remains an unresolved challenge in oil and gas engineering.In this paper,by utilizing the self-developed dynamic plugging apparatus and synthetic cores containing large-scale fractures,experimental research on the circulation plugging of different materials was conducted.Based on the D90 rule and fracture mechanical aperture model,we analyze the location of plugging layer under dynamic plugging mechanism.By setting different parameters of fracture width and injection pressure,the laws of cyclic plugging time,pressure bearing capacity and plugging layers formation were investigated.The results show that the comprehensive analysis of particle size and fracture aperture provides an accurate judgment of the entrance-plugging phenomenon.The bridging of solid materials in the leakage channel is a gradual process,and the formation of a stable plug requires 2–3 plug-leakage cycles.The first and second cyclic plugging time was positively correlated with the fracture width.Different scales of fractures were successfully plugged with the bearing pressure greater than 6 MPa,but there were significant differences in the composition of the plugging layer.The experimental results can effectively prove that the utilized plugging agent is effective and provides an effective reference for dynamic plugging operation.
基金National Natural Science Foundation of China,No.42171255,No.41971216。
文摘The application of ecosystem services(ES)theories in land consolidation is a confusing issue that has long plagued scholars and government officials.As the upgraded version of traditional land consolidation,comprehensive land consolidation(CLC)emphasizes ecological benefits,but it does not achieve the expected effect during the pilot phase.This study first proposed a theoretical analysis framework based on ES knowledge to answer the three key questions of why,where,and how to implement CLC better.Taking mountainous counties as the study area,we found that ES trade-offs/synergies,bundles,and drivers were significantly affected by scale effects.ES knowledge can play a crucial role in designing multi-scale CLC strategies regarding the objective,zoning,intensity,and mode.Specifically,mitigating the significant trade-offs between recreational opportunities,food production,and other ES is the top priority of CLC.Land consolidation zoning based on the ES bundles analysis is more rational and can provide the scientific premise for designing locally adapted CLC measures.Land consolidation can be classified into high-intensity direct intervention and low-intensity indirect intervention modes,based on the major drivers of ES.These findings help narrow the gap between ES and CLC practices.
基金funded by the Ministry of Public Security Science and Technology Program Project(No.2023LL35)the Key Laboratory of Smart Policing and National Security Risk Governance,Sichuan Province(No.ZHZZZD2302).
文摘As the use of deepfake facial videos proliferate,the associated threats to social security and integrity cannot be overstated.Effective methods for detecting forged facial videos are thus urgently needed.While many deep learning-based facial forgery detection approaches show promise,they often fail to delve deeply into the complex relationships between image features and forgery indicators,limiting their effectiveness to specific forgery techniques.To address this challenge,we propose a dual-branch collaborative deepfake detection network.The network processes video frame images as input,where a specialized noise extraction module initially extracts the noise feature maps.Subsequently,the original facial images and corresponding noise maps are directed into two parallel feature extraction branches to concurrently learn texture and noise forgery clues.An attention mechanism is employed between the two branches to facilitate mutual guidance and enhancement of texture and noise features across four different scales.This dual-modal feature integration enhances sensitivity to forgery artifacts and boosts generalization ability across various forgery techniques.Features from both branches are then effectively combined and processed through a multi-layer perception layer to distinguish between real and forged video.Experimental results on benchmark deepfake detection datasets demonstrate that our approach outperforms existing state-of-the-art methods in terms of detection performance,accuracy,and generalization ability.
文摘Objective To report the development,validation,and findings of the Multi-dimensional Attention Rating Scale(MARS),a self-report tool crafted to evaluate six-dimension attention levels.Methods The MARS was developed based on Classical Test Theory(CTT).Totally 202 highly educated healthy adult participants were recruited for reliability and validity tests.Reliability was measured using Cronbach's alpha and test-retest reliability.Structural validity was explored using principal component analysis.Criterion validity was analyzed by correlating MARS scores with the Toronto Hospital Alertness Test(THAT),the Attentional Control Scale(ACS),and the Attention Network Test(ANT).Results The MARS comprises 12 items spanning six distinct dimensions of attention:focused attention,sustained attention,shifting attention,selective attention,divided attention,and response inhibition.As assessed by six experts,the content validation index(CVI)was 0.95,the Cronbach's alpha for the MARS was 0.78,and the test-retest reliability was 0.81.Four factors were identified(cumulative variance contribution rate 68.79%).The total score of MARS was correlated positively with THAT(r=0.60,P<0.01)and ACS(r=0.78,P<0.01)and negatively with ANT's reaction time for alerting(r=−0.31,P=0.049).Conclusion The MARS can reliably and validly assess six-dimension attention levels in real-world settings and is expected to be a new tool for assessing multi-dimensional attention impairments in different mental disorders.
基金supported by the National Natural Science Foundation of China(Grant No.U24A20580)National Natural Science Foundation of China(Grant No.42171298)+1 种基金Natural Science Foundation of Chongqing,China(Grant No.CSTB2023NSCQLZX0009)Philosophy and Social Science Major Project of Chongqing Municipal Education Commission(Grant No.24SKZDZX04).
文摘Urban sprawl is a critical challenge in the urban development trajectory of developing countries,necessitating precise measurement,trend projection,and strategic management to achieve sustainable urban growth.This study focuses on the Yangtze River Economic Belt(YREB)as a case region and introduces a comprehensive evaluation framework that incorporates multidimensional factors and addresses the scale effects of urban sprawl.We emphasize the value of a systematic geographical approach by quantifying urban sprawl through simulated scenarios and analyzing its driving factors.We constructed an innovative urban sprawl index(USI)to assess the degree of sprawl within the YREB.This assessment integrates two geographic models with an artificial neural network algorithm,enabling simulation of urban sprawl trends under two future scenarios for 2035.Additionally,two analytical methods were employed to identify the key driving mechanisms of urban sprawl in the region.Findings indicate a strong correlation between urban scale and the extent of urban sprawl:larger urban areas exhibit more pronounced sprawl,with agglomeration and morphological transformations identified as primary contributors to urban sprawl.The study further reveals an intricate association between urban sprawl and the compactness of urban internal structures.While both development scenarios offer distinct advantages,the Coordinated Development Scenario is projected to foster a more balanced urban expansion.The robustness of the evaluation framework was enhanced through simulation and an in-depth analysis of internal mechanisms,bolstering confidence in its applicability.We advocate for the adoption and continued refinement of this framework as a tool for promoting balanced urban growth.The strategic recommendations provided herein are vital for mitigating multi-scale urban sprawl,advancing economic development,and improving residents’quality of life across cities in the YREB.
基金financially supported by the Natural Science Foundation of Shandong(ZR2023ME014)。
文摘Electrocatalysis has been extensively explored for the storage and conversion of renewable electric power.Understanding the physisorption and chemisorption processes at electrified solid–liquid interfaces(ESLIs)is crucial for revealing the typical surface restructuring and catalyst dissolution during electrocatalysis.Although advanced in situ tools and theoretical models have been proposed[1,2],identifying the nature of the active sites with atomic-scale spatial resolution remains a challenge,especially at ESLIs.In a recent work published in Nature,Zhang et al.[3]reported a groundbreaking atomic-resolution imaging of the structural dynamics of Cu nanowire catalysts in ESLIs for electrochemical CO_(2)reduction(ECR).
基金supported by the confidential research grant No.a8317。
文摘To address the issues of frequent identity switches(IDs)and degraded identification accuracy in multi object tracking(MOT)under complex occlusion scenarios,this study proposes an occlusion-robust tracking framework based on face-pedestrian joint feature modeling.By constructing a joint tracking model centered on“intra-class independent tracking+cross-category dynamic binding”,designing a multi-modal matching metric with spatio-temporal and appearance constraints,and innovatively introducing a cross-category feature mutual verification mechanism and a dual matching strategy,this work effectively resolves performance degradation in traditional single-category tracking methods caused by short-term occlusion,cross-camera tracking,and crowded environments.Experiments on the Chokepoint_Face_Pedestrian_Track test set demonstrate that in complex scenes,the proposed method improves Face-Pedestrian Matching F1 area under the curve(F1 AUC)by approximately 4 to 43 percentage points compared to several traditional methods.The joint tracking model achieves overall performance metrics of IDF1:85.1825%and MOTA:86.5956%,representing improvements of 0.91 and 0.06 percentage points,respectively,over the baseline model.Ablation studies confirm the effectiveness of key modules such as the Intersection over Area(IoA)/Intersection over Union(IoU)joint metric and dynamic threshold adjustment,validating the significant role of the cross-category identity matching mechanism in enhancing tracking stability.Our_model shows a 16.7%frame per second(FPS)drop vs.fairness of detection and re-identification in multiple object tracking(FairMOT),with its cross-category binding module adding aboute 10%overhead,yet maintains near-real-time performance for essential face-pedestrian tracking at small resolutions.
文摘High-throughput transcriptomics has evolved from bulk RNA-seq to single-cell and spatial profiling,yet its clinical translation still depends on effective integration across diverse omics and data modalities.Emerging foundation models and multimodal learning frameworks are enabling scalable and transferable representations of cellular states,while advances in interpretability and real-world data integration are bridging the gap between discovery and clinical application.This paper outlines a concise roadmap for AI-driven,transcriptome-centered multi-omics integration in precision medicine(Figure 1).
