AIM:To investigate the effects of binocular fusional C-optotypes(positive/negative)and 2D planar C-optotypes on the amplitude and stability of transient accommodation(TAC)in adults,and to provide a basis for non-conta...AIM:To investigate the effects of binocular fusional C-optotypes(positive/negative)and 2D planar C-optotypes on the amplitude and stability of transient accommodation(TAC)in adults,and to provide a basis for non-contact myopia intervention.METHODS:This was a self-controlled study.Using redblue 3D technology,four experimental stages were set up:Test A[fixating on the 1 m negative fusional C-optotypes,8△base-in(BI)],Test B(fixating on the 5 m planar C-optotypes),Test C(fixating on the 1 m planar C-optotypes),and Test D[fixating on the 1 m positive fusional C-optotypes,20△base-out(BO)].A WAM-5500 open-field autorefractor was used to measure TAC and accommodative microfluctuations[evaluated via interquartile range(IQR)and median-based coefficient of variation(CVmed)].Additionally,the convergence accommodation to convergence(CA/C)ratio was calculated,and a visual fatigue questionnaire was administered to assess participants’subjective visual comfort.RESULTS:A total of 21 subjects(7 males,14 females;aged 23-41y)with normal binocular visual function were enrolled.The results showed that the TAC increased gradually across the four stages,and these values were Test A(-0.35±0.26 D)<Test B(-0.46±0.24 D)<Test C(-0.77±0.32 D)<Test D(-1.38±0.31 D).There were significant overall differences(F=56.136,P<0.001).Compared with Test C,Test A reduced TAC by 0.42 D(P<0.05),while Test D increased it by 0.61 D(P<0.001).There was no significant intergroup difference in accommodative fluctuation amplitude(all P>0.05),but the fluctuation stability of Test D showed a significant difference between the first 20s and the second 20s(P=0.017).The CA/C ratio was significantly higher in Test D(0.05±0.02 D/△)than in Test A(0.03±0.02 D/△,P=0.007),indicating stronger accommodation-convergence linkage during positive fusional fixation.The visual fatigue scores of all stages were low(median 0-1),with Test D slightly higher than Test B and Test C(P<0.05).No linear correlation was found between TAC and age(all r<0.1,P>0.05).CONCLUSION:Negative fusional C-optotypes induce ciliary muscle relaxation to reduce TAC,while positive fusional C-optotypes enhance accommodation-convergence coordination to increase TAC.The red-blue 3D-based noncontact training mode exhibits good safety(median visual fatigue scores:0-1 across all tests)and provides a novel dual-directional(relaxation-activation)strategy for myopia prevention and control.展开更多
This paper presents a reality-virtual fusional campus environment.It is an online 3D platform with some aspects of real information merged together.The whole platform is based on OpenSimulator with detailed geo-models...This paper presents a reality-virtual fusional campus environment.It is an online 3D platform with some aspects of real information merged together.The whole platform is based on OpenSimulator with detailed geo-models to represent the university campus.Some preliminary experiments were done to integrate the realistic information with the virtual campus for making the geo-environment not only with detailed indoor and outdoor models,but also with the real representations of the physical world.The overall motivation is to provide a framework with strong support for reality-virtuality fusional modeling in a collaborative 3D online platform for research purposes.展开更多
AIM:To evaluate the differences in near point of convergence(NPC),fusional vergence,saccadic eye movements,versional eye movements,and heterophoria between patients diagnosed with Parkinson’s disease(PD)and healthy s...AIM:To evaluate the differences in near point of convergence(NPC),fusional vergence,saccadic eye movements,versional eye movements,and heterophoria between patients diagnosed with Parkinson’s disease(PD)and healthy subjects.METHODS:A cross-sectional comparative study was conducted,enrolling two cohorts:a PD group and a healthy control group.The PD group was recruited via non-random convenience sampling,while the control group was selected randomly from individuals without PD.All participants were screened according to predefined inclusion and exclusion criteria before undergoing a comprehensive optometric assessment,which included measurements of uncorrected visual acuity,corrected visual acuity,and objective and subjective refraction.Subsequently,binocular vision function evaluations were performed,covering NPC measurement,fusional vergence reserve assessment at both distance and near,saccadic eye movement testing,and versional eye movement and heterophoria assessment.RESULTS:A total of 42 PD patients and 41 healthy controls were included in the final analysis.The two groups were well-matched in terms of sex distribution[29 males(69.0%)in the PD group vs 29 males(70.7%)in the control group,P=0.867]and mean age(55.3±9.6y in the PD group vs 54.9±9.8y in the control group,P=0.866).The prevalence of abnormal versional eye movements was significantly higher in the PD group than in the control group(23.81%,95%CI:12.05%-39.45%vs 7.32%,95%CI:1.54%-19.92%;P=0.025).Near exophoria was more prevalent in PD patients(61.90%,95%CI:45.64%-76.43%)than in controls(17.07%,95%CI:7.15%-32.06%),with a significant difference[odds ratio(OR)=7.99;95%CI:2.83-21.99;P<0.001].The mean NPC was significantly greater(more receded)in the PD group than in the control group(9.01±3.74 cm vs 7.20±2.15 cm;P=0.007).A statistically significant positive correlation was observed between PD severity and NPC values(Pearson’s correlation coefficient=0.309;P=0.046).Except for distance baseout break and distance base-out recovery values,all other fusional vergence parameters were significantly lower in the PD group than in the control group(P<0.05).The mean saccadic test score was significantly lower in PD patients than in controls(3.29±0.57 vs 3.78±0.42;P<0.001).Among all fusional vergence indices,near base-in blur yielded the highest area under the curve(AUC=0.877),with a sensitivity of 69%and specificity of 90%,followed by distance base-out blur(AUC=0.824,sensitivity=97.6%,specificity=66.7%),near base-out blur(AUC=0.814,sensitivity=76.2%,specificity=72.7%),near base-out break(AUC=0.749,sensitivity=78.6%,specificity=67.6%),and near base-out recovery(AUC=0.749,sensitivity=95.2%,specificity=50%).CONCLUSION:PD is associated with significant binocular vision function impairment,with receded NPC and reduced near fusional vergence reserves being the most prominent disorders.These findings highlight the potential value of binocular vision assessment as a non-invasive biomarker for the early detection and clinical monitoring of PD.展开更多
Visible and infrared(RGB-IR)fusion object detection plays an important role in security,disaster relief,etc.In recent years,deep-learning-based RGB-IR fusion detection methods have been developing rapidly,but still st...Visible and infrared(RGB-IR)fusion object detection plays an important role in security,disaster relief,etc.In recent years,deep-learning-based RGB-IR fusion detection methods have been developing rapidly,but still struggle to deal with the complex and changing scenarios captured by drones,mainly due to two reasons:(A)RGB-IR fusion detectors are susceptible to inferior inputs that degrade performance and stability.(B)RGB-IR fusion detectors are susceptible to redundant features that reduce accuracy and efficiency.In this paper,an innovative RGB-IR fusion detection framework based on global-local feature optimization,named GLFDet,is proposed to improve the detection performance and efficiency of drone-captured objects.The key components of GLFDet include a Global Feature Optimization(GFO)module,a Local Feature Optimization(LFO)module and a Channel Separation Fusion(CSF)module.Specifically,GFO calculates the information content of the input image from the frequency domain and optimizes the features holistically.Then,LFO dynamically selects high-value features and filters out low-value features before fusion,which significantly improves the efficiency of fusion.Finally,CSF fuses the RGB and IR features across the corresponding channels,which avoids the rearrangement of the channel relationships and enhances the model stability.Extensive experimental results show that the proposed method achieves the best performance on three popular RGB-IR datasets Drone Vehicle,VEDAI,and LLVIP.In addition,GLFDet is more lightweight than other comparable models,making it more appealing to edge devices such as drones.The code is available at https://github.com/lao chen330/GLFDet.展开更多
A dual-phase synergistic enhancement method was adopted to strengthen the Al-Mn-Mg-Sc-Zr alloy fabricated by laser powder bed fusion(LPBF)by leveraging the unique advantages of Er and TiB_(2).Spherical powders of 0.5w...A dual-phase synergistic enhancement method was adopted to strengthen the Al-Mn-Mg-Sc-Zr alloy fabricated by laser powder bed fusion(LPBF)by leveraging the unique advantages of Er and TiB_(2).Spherical powders of 0.5wt%Er-1wt%TiB_(2)/Al-Mn-Mg-Sc-Zr nanocomposite were prepared using vacuum homogenization technique,and the density of samples prepared through the LPBF process reached 99.8%.The strengthening and toughening mechanisms of Er-TiB_(2)were investigated.The results show that Al_(3)Er diffraction peaks are detected by X-ray diffraction analysis,and texture strength decreases according to electron backscatter diffraction results.The added Er and TiB_(2)nano-reinforcing phases act as heterogeneous nucleation sites during the LPBF forming process,hindering grain growth and effectively refining the grains.After incorporating the Er-TiB_(2)dual-phase nano-reinforcing phases,the tensile strength and elongation at break of the LPBF-deposited samples reach 550 MPa and 18.7%,which are 13.4%and 26.4%higher than those of the matrix material,respectively.展开更多
Copper manufactured by laser powder bed fusion(LPBF)process typically exhibits poor strength-ductility coordination,and the addition of strengthening phases is an effective way to address this issue.To explore the eff...Copper manufactured by laser powder bed fusion(LPBF)process typically exhibits poor strength-ductility coordination,and the addition of strengthening phases is an effective way to address this issue.To explore the effects of strengthening phases on Cu,Cu-carbon nanotubes(CNTs)composites were prepared using LPBF technique with Cu-CNTs mixed powder as the matrix.The formability,microstructure,mechanical properties,electrical conductivity,and thermal properties were studied.The result shows that the prepared composites have high relative density.The addition of CNTs results in inhomogeneous equiaxed grains at the edges of the molten pool and columnar grains at the center.Compared with pure copper,the overall mechanical properties of the composite are improved:tensile strength increases by 52.8%and elongation increases by 146.4%;the electrical and thermal properties are also enhanced:thermal conductivity increases by 10.8%and electrical conductivity increases by 12.7%.展开更多
Laser powder bed fusion(LPBF)is highly suitable for forming 18Ni300 mold steel,thanks to its excellent capability in manufacturing complex shapes and outstanding capacity for regulating microstructures.It is widely us...Laser powder bed fusion(LPBF)is highly suitable for forming 18Ni300 mold steel,thanks to its excellent capability in manufacturing complex shapes and outstanding capacity for regulating microstructures.It is widely used in fields such as injection molding,die casting,and stamping dies.Adding reinforcing particles into steel is an effective means to improve its performance.Nb/18Ni300 composites were fabricated by LPBF using two kinds of Nb powders with different particle sizes,and their microstructures and properties were studied.The results show that the unmelted Nb particles are uniformly distributed in the 18Ni300 matrix and the grains are refined,which is particularly pronounced with fine Nb particles.In addition,element diffusion occurs between the particles and the matrix.The main phases of the base alloy are α-Fe and a small amount of γ-Fe.With the addition of Nb,part of the α-Fe is transformed into γ-Fe,and unmelted Nb phases appear.The addition of Nb also enhances the hardness and wear resistance of the composites but slightly reduces their tensile properties.After aging treatment,the molten pools and grain boundaries become blurred,grains are further refined,and the interfaces around the particles are thinned.The aging treatment also promotes the formation of reverted austenite.The hardness,ultimate tensile strength,and volumetric wear rate of the base alloy reach 51.9 HRC,1704 MPa,and 17.8×10^(-6) mm^(3)/(N·m),respectively.In contrast,the sample added with fine Nb particles has the highest hardness(56.1 HRC),ultimate tensile strength(1892 MPa)and yield strength(1842 MPa),and the volume wear rate of the sample added with coarse Nb particles is reduced by 90%to 1.7×10^(-6) mm^(3)/(N·m).展开更多
Parkinson’s disease remains a major clinical issue in terms of early detection,especially during its prodromal stage when symptoms are not evident or not distinct.To address this problem,we proposed a new deep learni...Parkinson’s disease remains a major clinical issue in terms of early detection,especially during its prodromal stage when symptoms are not evident or not distinct.To address this problem,we proposed a new deep learning 2-based approach for detecting Parkinson’s disease before any of the overt symptoms develop during their prodromal stage.We used 5 publicly accessible datasets,including UCI Parkinson’s Voice,Spiral Drawings,PaHaW,NewHandPD,and PPMI,and implemented a dual stream CNN–BiLSTM architecture with Fisher-weighted feature merging and SHAP-based explanation.The findings reveal that the model’s performance was superior and achieved 98.2%,a F1-score of 0.981,and AUC of 0.991 on the UCI Voice dataset.The model’s performance on the remaining datasets was also comparable,with up to a 2–7 percent betterment in accuracy compared to existing strong models such as CNN–RNN–MLP,ILN–GNet,and CASENet.Across the evidence,the findings back the diagnostic promise of micro-tremor assessment and demonstrate that combining temporal and spatial features with a scatter-based segment for a multi-modal approach can be an effective and scalable platform for an“early,”interpretable PD screening system.展开更多
Fault diagnosis of rolling bearings is crucial for ensuring the stable operation of mechanical equipment and production safety in industrial environments.However,due to the nonlinearity and non-stationarity of collect...Fault diagnosis of rolling bearings is crucial for ensuring the stable operation of mechanical equipment and production safety in industrial environments.However,due to the nonlinearity and non-stationarity of collected vibration signals,single-modal methods struggle to capture fault features fully.This paper proposes a rolling bearing fault diagnosis method based on multi-modal information fusion.The method first employs the Hippopotamus Optimization Algorithm(HO)to optimize the number of modes in Variational Mode Decomposition(VMD)to achieve optimal modal decomposition performance.It combines Convolutional Neural Networks(CNN)and Gated Recurrent Units(GRU)to extract temporal features from one-dimensional time-series signals.Meanwhile,the Markovian Transition Field(MTF)is used to transform one-dimensional signals into two-dimensional images for spatial feature mining.Through visualization techniques,the effectiveness of generated images from different parameter combinations is compared to determine the optimal parameter configuration.A multi-modal network(GSTCN)is constructed by integrating Swin-Transformer and the Convolutional Block Attention Module(CBAM),where the attention module is utilized to enhance fault features.Finally,the fault features extracted from different modalities are deeply fused and fed into a fully connected layer to complete fault classification.Experimental results show that the GSTCN model achieves an average diagnostic accuracy of 99.5%across three datasets,significantly outperforming existing comparison methods.This demonstrates that the proposed model has high diagnostic precision and good generalization ability,providing an efficient and reliable solution for rolling bearing fault diagnosis.展开更多
The process of nuclear fusion in the presence of a laser field was theoretically analyzed.The analysis is applicable to most fusion reactions and different types of currently available intense lasers,from X-ray free-e...The process of nuclear fusion in the presence of a laser field was theoretically analyzed.The analysis is applicable to most fusion reactions and different types of currently available intense lasers,from X-ray free-electron lasers to solid-state near-infrared lasers.Laser fields were shown to enhance the fusion yields,and the mechanism of this enhancement was explained.Low-frequency lasers are more efficient in enhancing fusion than high-frequency lasers.The calculation results show enhancements of fusion yields by orders of magnitude with currently available intense low-frequency laser fields.The temperature requirement for controlled nuclear fusion may be reduced with the aid of intense laser fields.展开更多
Unmanned aerial vehicles(UAVs),especially quadcopters,have become indispensable in numerous industrial and scientific applications due to their flexibility,lowcost,and capability to operate in dynamic environments.Thi...Unmanned aerial vehicles(UAVs),especially quadcopters,have become indispensable in numerous industrial and scientific applications due to their flexibility,lowcost,and capability to operate in dynamic environments.This paper presents a complete design and implementation of a compact autonomous quadcopter capable of trajectory tracking,object detection,precision landing,and real-time telemetry via long-range communication protocols.The system integrates an onboard flight controller running real-time sensor fusion algorithms,a vision-based detection system on a companion single-board computer,and a telemetry unit using Long Range(LoRa)communication.Extensive flight tests were conducted to validate the system’s stability,communication range,and autonomous capabilities.Potential applications in law enforcement,agriculture,search and rescue,and environmental monitoring are also discussed.展开更多
The increasing complexity of intelligent sensing environments,driven by the growth of Internet of Things technologies,has created a strong demand for neuromorphic systems capable of real-time,low-power multisensory pe...The increasing complexity of intelligent sensing environments,driven by the growth of Internet of Things technologies,has created a strong demand for neuromorphic systems capable of real-time,low-power multisensory perception.Traditional sensory architectures,constrained by single-modal processing and centralized computing,struggle to meet the requirements of diverse and dynamic input conditions.Multisensory neuromorphic devices offer a promising solution by mimicking the distributed,event-driven processing of biological systems.Recent efforts have explored synaptic devices and material systems that respond to various input modalities,including visual,tactile,thermal,and chemical stimuli.However,challenges remain in signal conversion,encoding compatibility,and the fusion of heterogeneous inputs without loss of unisensory information.This review provides a comprehensive overview of the physical mechanisms,device behaviors,and integration strategies that underpin signal processing in neuromorphic hardware.We highlight synaptic mechanisms conducive to cross-modal interaction,analyze representative signal fusion approaches at the device level,and discuss future directions for constructing efficient,scalable,and biologically inspired multisensory neuromorphic systems.展开更多
Traffic sign detection is a critical component of driving systems.Single-stage network-based traffic sign detection algorithms,renowned for their fast detection speeds and high accuracy,have become the dominant approa...Traffic sign detection is a critical component of driving systems.Single-stage network-based traffic sign detection algorithms,renowned for their fast detection speeds and high accuracy,have become the dominant approach in current practices.However,in complex and dynamic traffic scenes,particularly with smaller traffic sign objects,challenges such as missed and false detections can lead to reduced overall detection accuracy.To address this issue,this paper proposes a detection algorithm that integrates edge and shape information.Recognizing that traffic signs have specific shapes and distinct edge contours,this paper introduces an edge feature extraction branch within the backbone network,enabling adaptive fusion with features of the same hierarchical level.Additionally,a shape prior convolution module is designed to replaces the first two convolutional modules of the backbone network,aimed at enhancing the model's perception ability for specific shape objects and reducing its sensitivity to background noise.The algorithm was evaluated on the CCTSDB and TT100k datasets,and compared to YOLOv8s,the mAP50 values increased by 3.0%and 10.4%,respectively,demonstrating the effectiveness of the proposed method in improving the accuracy of traffic sign detection.展开更多
Tungsten(W),owing to its exceptional physical and chemical properties,is a promising candidate for plasma-facing materials.However,its intrinsic brittleness makes it highly susceptible to cracking during its processin...Tungsten(W),owing to its exceptional physical and chemical properties,is a promising candidate for plasma-facing materials.However,its intrinsic brittleness makes it highly susceptible to cracking during its processing,especially using the laser powder bed fusion(LPBF)process with high energy input.Alloying has been recognized as an effective strategy to deal with this challenge,yet investigations into rare earth element alloying for tungsten and corresponding LPBF production remain limited.In this study,yttrium(Y),a rare earth element,was introduced to alloy with tungsten,and W-Y alloys were fabricated via LPBF at various laser energy densities.Finite element simulations were conducted to predict the temperature field distributions of W-Y alloy under different laser energy densities,providing insight into the formation of metallurgical defects at various laser energy inputs.It revealed that at a suitable laser energy density of 500 J/mm^(3),the fabricated W-Y specimens exhibited smooth and flat melt paths without discernible internal pores or cracks,achieving densification of 99.3%.The W-Y alloy had a refined microstructure with fine columnar and equiaxial grains,with an average grain size of 15.83μm.The compressive strength and elongation after fracture of the W-Y alloy were 1531.93 MPa and 21.57%,respectively.The excellent hardness of 520 HV_(0.2)and wear resistance with a coefficient of friction(COF)of 0.47 were obtained.The enhanced mechanical performance could be attributed to grain refinement strengthening and dislocation strengthening.The study on the tensile properties of the intrinsic brittle W-Y sample was conducted for the first time,highlighting the intrinsic challenge of improving the tensile ductility of tungsten.This study not only provides a new perspective on rare-earth-alloyed tungsten but also offers a scientific reference for LPBF processing of high-performance W-Y alloy.展开更多
Autonomous vehicles rely heavily on accurate and efficient scene segmentation for safe navigation and efficient operations.Traditional Bird’s Eye View(BEV)methods on semantic scene segmentation,which leverage multimo...Autonomous vehicles rely heavily on accurate and efficient scene segmentation for safe navigation and efficient operations.Traditional Bird’s Eye View(BEV)methods on semantic scene segmentation,which leverage multimodal sensor fusion,often struggle with noisy data and demand high-performance GPUs,leading to sensor misalignment and performance degradation.This paper introduces an Enhanced Channel Attention BEV(ECABEV),a novel approach designed to address the challenges under insufficient GPU memory conditions.ECABEV integrates camera and radar data through a de-noise enhanced channel attention mechanism,which utilizes global average and max pooling to effectively filter out noise while preserving discriminative features.Furthermore,an improved fusion approach is proposed to efficiently merge categorical data across modalities.To reduce computational overhead,a bilinear interpolation layer normalizationmethod is devised to ensure spatial feature fidelity.Moreover,a scalable crossentropy loss function is further designed to handle the imbalanced classes with less computational efficiency sacrifice.Extensive experiments on the nuScenes dataset demonstrate that ECABEV achieves state-of-the-art performance with an IoU of 39.961,using a lightweight ViT-B/14 backbone and lower resolution(224×224).Our approach highlights its cost-effectiveness and practical applicability,even on low-end devices.The code is publicly available at:https://github.com/YYF-CQU/ECABEV.git.展开更多
The initial noise present in the depth images obtained with RGB-D sensors is a combination of hardware limitations in addition to the environmental factors,due to the limited capabilities of sensors,which also produce...The initial noise present in the depth images obtained with RGB-D sensors is a combination of hardware limitations in addition to the environmental factors,due to the limited capabilities of sensors,which also produce poor computer vision results.The common image denoising techniques tend to remove significant image details and also remove noise,provided they are based on space and frequency filtering.The updated framework presented in this paper is a novel denoising model that makes use of Boruta-driven feature selection using a Long Short-Term Memory Autoencoder(LSTMAE).The Boruta algorithm identifies the most useful depth features that are used to maximize the spatial structure integrity and reduce redundancy.An LSTMAE is then used to process these selected features and model depth pixel sequences to generate robust,noise-resistant representations.The system uses the encoder to encode the input data into a latent space that has been compressed before it is decoded to retrieve the clean image.Experiments on a benchmark data set show that the suggested technique attains a PSNR of 45 dB and an SSIM of 0.90,which is 10 dB higher than the performance of conventional convolutional autoencoders and 15 times higher than that of the wavelet-based models.Moreover,the feature selection step will decrease the input dimensionality by 40%,resulting in a 37.5%reduction in training time and a real-time inference rate of 200 FPS.Boruta-LSTMAE framework,therefore,offers a highly efficient and scalable system for depth image denoising,with a high potential to be applied to close-range 3D systems,such as robotic manipulation and gesture-based interfaces.展开更多
Small object detection has been a focus of attention since the emergence of deep learning-based object detection.Although classical object detection frameworks have made significant contributions to the development of...Small object detection has been a focus of attention since the emergence of deep learning-based object detection.Although classical object detection frameworks have made significant contributions to the development of object detection,there are still many issues to be resolved in detecting small objects due to the inherent complexity and diversity of real-world visual scenes.In particular,the YOLO(You Only Look Once)series of detection models,renowned for their real-time performance,have undergone numerous adaptations aimed at improving the detection of small targets.In this survey,we summarize the state-of-the-art YOLO-based small object detection methods.This review presents a systematic categorization of YOLO-based approaches for small-object detection,organized into four methodological avenues,namely attention-based feature enhancement,detection-head optimization,loss function,and multi-scale feature fusion strategies.We then examine the principal challenges addressed by each category.Finally,we analyze the performance of thesemethods on public benchmarks and,by comparing current approaches,identify limitations and outline directions for future research.展开更多
At inference time,deep neural networks are susceptible to backdoor attacks,which can produce attackercontrolled outputs when inputs contain carefully crafted triggers.Existing defense methods often focus on specific a...At inference time,deep neural networks are susceptible to backdoor attacks,which can produce attackercontrolled outputs when inputs contain carefully crafted triggers.Existing defense methods often focus on specific attack types or incur high costs,such as data cleaning or model fine-tuning.In contrast,we argue that it is possible to achieve effective and generalizable defense without removing triggers or incurring high model-cleaning costs.Fromthe attacker’s perspective and based on characteristics of vulnerable neuron activation anomalies,we propose an Adaptive Feature Injection(AFI)method for black-box backdoor detection.AFI employs a pre-trained image encoder to extract multi-level deep features and constructs a dynamic weight fusionmechanism for precise identification and interception of poisoned samples.Specifically,we select the control samples with the largest feature differences fromthe clean dataset via feature-space analysis,and generate blended sample pairs with the test sample using dynamic linear interpolation.The detection statistic is computed by measuring the divergence G(x)in model output responses.We systematically evaluate the effectiveness of AFI against representative backdoor attacks,including BadNets,Blend,WaNet,and IAB,on three benchmark datasets:MNIST,CIFAR-10,and ImageNet.Experimental results show that AFI can effectively detect poisoned samples,achieving average detection rates of 95.20%,94.15%,and 86.49%on these datasets,respectively.Compared with existing methods,AFI demonstrates strong cross-domain generalization ability and robustness to unknown attacks.展开更多
Deep learning has made significant progress in the field of oriented object detection for remote sensing images.However,existing methods still face challenges when dealing with difficult tasks such as multi-scale targ...Deep learning has made significant progress in the field of oriented object detection for remote sensing images.However,existing methods still face challenges when dealing with difficult tasks such as multi-scale targets,complex backgrounds,and small objects in remote sensing.Maintaining model lightweight to address resource constraints in remote sensing scenarios while improving task completion for remote sensing tasks remains a research hotspot.Therefore,we propose an enhanced multi-scale feature extraction lightweight network EM-YOLO based on the YOLOv8s architecture,specifically optimized for the characteristics of large target scale variations,diverse orientations,and numerous small objects in remote sensing images.Our innovations lie in two main aspects:First,a dynamic snake convolution(DSC)is introduced into the backbone network to enhance the model’s feature extraction capability for oriented targets.Second,an innovative focusing-diffusion module is designed in the feature fusion neck to effectively integrate multi-scale feature information.Finally,we introduce Layer-Adaptive Sparsity for magnitude-based Pruning(LASP)method to perform lightweight network pruning to better complete tasks in resource-constrained scenarios.Experimental results on the lightweight platform Orin demonstrate that the proposed method significantly outperforms the original YOLOv8s model in oriented remote sensing object detection tasks,and achieves comparable or superior performance to state-of-the-art methods on three authoritative remote sensing datasets(DOTA v1.0,DOTA v1.5,and HRSC2016).展开更多
文摘AIM:To investigate the effects of binocular fusional C-optotypes(positive/negative)and 2D planar C-optotypes on the amplitude and stability of transient accommodation(TAC)in adults,and to provide a basis for non-contact myopia intervention.METHODS:This was a self-controlled study.Using redblue 3D technology,four experimental stages were set up:Test A[fixating on the 1 m negative fusional C-optotypes,8△base-in(BI)],Test B(fixating on the 5 m planar C-optotypes),Test C(fixating on the 1 m planar C-optotypes),and Test D[fixating on the 1 m positive fusional C-optotypes,20△base-out(BO)].A WAM-5500 open-field autorefractor was used to measure TAC and accommodative microfluctuations[evaluated via interquartile range(IQR)and median-based coefficient of variation(CVmed)].Additionally,the convergence accommodation to convergence(CA/C)ratio was calculated,and a visual fatigue questionnaire was administered to assess participants’subjective visual comfort.RESULTS:A total of 21 subjects(7 males,14 females;aged 23-41y)with normal binocular visual function were enrolled.The results showed that the TAC increased gradually across the four stages,and these values were Test A(-0.35±0.26 D)<Test B(-0.46±0.24 D)<Test C(-0.77±0.32 D)<Test D(-1.38±0.31 D).There were significant overall differences(F=56.136,P<0.001).Compared with Test C,Test A reduced TAC by 0.42 D(P<0.05),while Test D increased it by 0.61 D(P<0.001).There was no significant intergroup difference in accommodative fluctuation amplitude(all P>0.05),but the fluctuation stability of Test D showed a significant difference between the first 20s and the second 20s(P=0.017).The CA/C ratio was significantly higher in Test D(0.05±0.02 D/△)than in Test A(0.03±0.02 D/△,P=0.007),indicating stronger accommodation-convergence linkage during positive fusional fixation.The visual fatigue scores of all stages were low(median 0-1),with Test D slightly higher than Test B and Test C(P<0.05).No linear correlation was found between TAC and age(all r<0.1,P>0.05).CONCLUSION:Negative fusional C-optotypes induce ciliary muscle relaxation to reduce TAC,while positive fusional C-optotypes enhance accommodation-convergence coordination to increase TAC.The red-blue 3D-based noncontact training mode exhibits good safety(median visual fatigue scores:0-1 across all tests)and provides a novel dual-directional(relaxation-activation)strategy for myopia prevention and control.
基金the Direct Grant of the Chinese University of Hong Kong (No.2021064)the National High Technology Research and Development Program of China (No.2010AA122202)
文摘This paper presents a reality-virtual fusional campus environment.It is an online 3D platform with some aspects of real information merged together.The whole platform is based on OpenSimulator with detailed geo-models to represent the university campus.Some preliminary experiments were done to integrate the realistic information with the virtual campus for making the geo-environment not only with detailed indoor and outdoor models,but also with the real representations of the physical world.The overall motivation is to provide a framework with strong support for reality-virtuality fusional modeling in a collaborative 3D online platform for research purposes.
基金Supported by Mashhad University of Medical Sciences.
文摘AIM:To evaluate the differences in near point of convergence(NPC),fusional vergence,saccadic eye movements,versional eye movements,and heterophoria between patients diagnosed with Parkinson’s disease(PD)and healthy subjects.METHODS:A cross-sectional comparative study was conducted,enrolling two cohorts:a PD group and a healthy control group.The PD group was recruited via non-random convenience sampling,while the control group was selected randomly from individuals without PD.All participants were screened according to predefined inclusion and exclusion criteria before undergoing a comprehensive optometric assessment,which included measurements of uncorrected visual acuity,corrected visual acuity,and objective and subjective refraction.Subsequently,binocular vision function evaluations were performed,covering NPC measurement,fusional vergence reserve assessment at both distance and near,saccadic eye movement testing,and versional eye movement and heterophoria assessment.RESULTS:A total of 42 PD patients and 41 healthy controls were included in the final analysis.The two groups were well-matched in terms of sex distribution[29 males(69.0%)in the PD group vs 29 males(70.7%)in the control group,P=0.867]and mean age(55.3±9.6y in the PD group vs 54.9±9.8y in the control group,P=0.866).The prevalence of abnormal versional eye movements was significantly higher in the PD group than in the control group(23.81%,95%CI:12.05%-39.45%vs 7.32%,95%CI:1.54%-19.92%;P=0.025).Near exophoria was more prevalent in PD patients(61.90%,95%CI:45.64%-76.43%)than in controls(17.07%,95%CI:7.15%-32.06%),with a significant difference[odds ratio(OR)=7.99;95%CI:2.83-21.99;P<0.001].The mean NPC was significantly greater(more receded)in the PD group than in the control group(9.01±3.74 cm vs 7.20±2.15 cm;P=0.007).A statistically significant positive correlation was observed between PD severity and NPC values(Pearson’s correlation coefficient=0.309;P=0.046).Except for distance baseout break and distance base-out recovery values,all other fusional vergence parameters were significantly lower in the PD group than in the control group(P<0.05).The mean saccadic test score was significantly lower in PD patients than in controls(3.29±0.57 vs 3.78±0.42;P<0.001).Among all fusional vergence indices,near base-in blur yielded the highest area under the curve(AUC=0.877),with a sensitivity of 69%and specificity of 90%,followed by distance base-out blur(AUC=0.824,sensitivity=97.6%,specificity=66.7%),near base-out blur(AUC=0.814,sensitivity=76.2%,specificity=72.7%),near base-out break(AUC=0.749,sensitivity=78.6%,specificity=67.6%),and near base-out recovery(AUC=0.749,sensitivity=95.2%,specificity=50%).CONCLUSION:PD is associated with significant binocular vision function impairment,with receded NPC and reduced near fusional vergence reserves being the most prominent disorders.These findings highlight the potential value of binocular vision assessment as a non-invasive biomarker for the early detection and clinical monitoring of PD.
基金supported by the National Natural Science Foundation of China(No.62276204)the Fundamental Research Funds for the Central Universities,China(No.YJSJ24011)+1 种基金the Natural Science Basic Research Program of Shaanxi,China(Nos.2022JM-340 and 2023-JC-QN-0710)the China Postdoctoral Science Foundation(Nos.2020T130494 and 2018M633470)。
文摘Visible and infrared(RGB-IR)fusion object detection plays an important role in security,disaster relief,etc.In recent years,deep-learning-based RGB-IR fusion detection methods have been developing rapidly,but still struggle to deal with the complex and changing scenarios captured by drones,mainly due to two reasons:(A)RGB-IR fusion detectors are susceptible to inferior inputs that degrade performance and stability.(B)RGB-IR fusion detectors are susceptible to redundant features that reduce accuracy and efficiency.In this paper,an innovative RGB-IR fusion detection framework based on global-local feature optimization,named GLFDet,is proposed to improve the detection performance and efficiency of drone-captured objects.The key components of GLFDet include a Global Feature Optimization(GFO)module,a Local Feature Optimization(LFO)module and a Channel Separation Fusion(CSF)module.Specifically,GFO calculates the information content of the input image from the frequency domain and optimizes the features holistically.Then,LFO dynamically selects high-value features and filters out low-value features before fusion,which significantly improves the efficiency of fusion.Finally,CSF fuses the RGB and IR features across the corresponding channels,which avoids the rearrangement of the channel relationships and enhances the model stability.Extensive experimental results show that the proposed method achieves the best performance on three popular RGB-IR datasets Drone Vehicle,VEDAI,and LLVIP.In addition,GLFDet is more lightweight than other comparable models,making it more appealing to edge devices such as drones.The code is available at https://github.com/lao chen330/GLFDet.
基金Shaanxi Province Qin Chuangyuan“Scientist+Engineer”Team Construction Project(2022KXJ-071)2022 Qin Chuangyuan Achievement Transformation Incubation Capacity Improvement Project(2022JH-ZHFHTS-0012)+8 种基金Shaanxi Province Key Research and Development Plan-“Two Chains”Integration Key Project-Qin Chuangyuan General Window Industrial Cluster Project(2023QCY-LL-02)Xixian New Area Science and Technology Plan(2022-YXYJ-003,2022-XXCY-010)2024 Scientific Research Project of Shaanxi National Defense Industry Vocational and Technical College(Gfy24-07)Shaanxi Vocational and Technical Education Association 2024 Vocational Education Teaching Reform Research Topic(2024SZX354)National Natural Science Foundation of China(U24A20115)2024 Shaanxi Provincial Education Department Service Local Special Scientific Research Program Project-Industrialization Cultivation Project(24JC005,24JC063)Shaanxi Province“14th Five-Year Plan”Education Science Plan,2024 Project(SGH24Y3181)National Key Research and Development Program of China(2023YFB4606400)Longmen Laboratory Frontier Exploration Topics Project(LMQYTSKT003)。
文摘A dual-phase synergistic enhancement method was adopted to strengthen the Al-Mn-Mg-Sc-Zr alloy fabricated by laser powder bed fusion(LPBF)by leveraging the unique advantages of Er and TiB_(2).Spherical powders of 0.5wt%Er-1wt%TiB_(2)/Al-Mn-Mg-Sc-Zr nanocomposite were prepared using vacuum homogenization technique,and the density of samples prepared through the LPBF process reached 99.8%.The strengthening and toughening mechanisms of Er-TiB_(2)were investigated.The results show that Al_(3)Er diffraction peaks are detected by X-ray diffraction analysis,and texture strength decreases according to electron backscatter diffraction results.The added Er and TiB_(2)nano-reinforcing phases act as heterogeneous nucleation sites during the LPBF forming process,hindering grain growth and effectively refining the grains.After incorporating the Er-TiB_(2)dual-phase nano-reinforcing phases,the tensile strength and elongation at break of the LPBF-deposited samples reach 550 MPa and 18.7%,which are 13.4%and 26.4%higher than those of the matrix material,respectively.
基金National Key Research and Development Program of China(2023YFB4606400)Supported by Longmen Laboratory Frontier Exploration Topics(LMQYTSKT003)。
文摘Copper manufactured by laser powder bed fusion(LPBF)process typically exhibits poor strength-ductility coordination,and the addition of strengthening phases is an effective way to address this issue.To explore the effects of strengthening phases on Cu,Cu-carbon nanotubes(CNTs)composites were prepared using LPBF technique with Cu-CNTs mixed powder as the matrix.The formability,microstructure,mechanical properties,electrical conductivity,and thermal properties were studied.The result shows that the prepared composites have high relative density.The addition of CNTs results in inhomogeneous equiaxed grains at the edges of the molten pool and columnar grains at the center.Compared with pure copper,the overall mechanical properties of the composite are improved:tensile strength increases by 52.8%and elongation increases by 146.4%;the electrical and thermal properties are also enhanced:thermal conductivity increases by 10.8%and electrical conductivity increases by 12.7%.
基金Key-Area Research and Development Program of Guangdong Province(2023B0909020004)Project of Innovation Research Team in Zhongshan(CXTD2023006)+1 种基金Natural Science Foundation of Guangdong Province(2023A1515011573)Zhongshan Social Welfare Science and Technology Research Project(2024B2022)。
文摘Laser powder bed fusion(LPBF)is highly suitable for forming 18Ni300 mold steel,thanks to its excellent capability in manufacturing complex shapes and outstanding capacity for regulating microstructures.It is widely used in fields such as injection molding,die casting,and stamping dies.Adding reinforcing particles into steel is an effective means to improve its performance.Nb/18Ni300 composites were fabricated by LPBF using two kinds of Nb powders with different particle sizes,and their microstructures and properties were studied.The results show that the unmelted Nb particles are uniformly distributed in the 18Ni300 matrix and the grains are refined,which is particularly pronounced with fine Nb particles.In addition,element diffusion occurs between the particles and the matrix.The main phases of the base alloy are α-Fe and a small amount of γ-Fe.With the addition of Nb,part of the α-Fe is transformed into γ-Fe,and unmelted Nb phases appear.The addition of Nb also enhances the hardness and wear resistance of the composites but slightly reduces their tensile properties.After aging treatment,the molten pools and grain boundaries become blurred,grains are further refined,and the interfaces around the particles are thinned.The aging treatment also promotes the formation of reverted austenite.The hardness,ultimate tensile strength,and volumetric wear rate of the base alloy reach 51.9 HRC,1704 MPa,and 17.8×10^(-6) mm^(3)/(N·m),respectively.In contrast,the sample added with fine Nb particles has the highest hardness(56.1 HRC),ultimate tensile strength(1892 MPa)and yield strength(1842 MPa),and the volume wear rate of the sample added with coarse Nb particles is reduced by 90%to 1.7×10^(-6) mm^(3)/(N·m).
基金supported via funding from Prince Sattam bin Abdulaziz University project number(PSAU/2025/03/32440).
文摘Parkinson’s disease remains a major clinical issue in terms of early detection,especially during its prodromal stage when symptoms are not evident or not distinct.To address this problem,we proposed a new deep learning 2-based approach for detecting Parkinson’s disease before any of the overt symptoms develop during their prodromal stage.We used 5 publicly accessible datasets,including UCI Parkinson’s Voice,Spiral Drawings,PaHaW,NewHandPD,and PPMI,and implemented a dual stream CNN–BiLSTM architecture with Fisher-weighted feature merging and SHAP-based explanation.The findings reveal that the model’s performance was superior and achieved 98.2%,a F1-score of 0.981,and AUC of 0.991 on the UCI Voice dataset.The model’s performance on the remaining datasets was also comparable,with up to a 2–7 percent betterment in accuracy compared to existing strong models such as CNN–RNN–MLP,ILN–GNet,and CASENet.Across the evidence,the findings back the diagnostic promise of micro-tremor assessment and demonstrate that combining temporal and spatial features with a scatter-based segment for a multi-modal approach can be an effective and scalable platform for an“early,”interpretable PD screening system.
基金funded by the Jilin Provincial Department of Science and Technology,grant number 20230101208JC.
文摘Fault diagnosis of rolling bearings is crucial for ensuring the stable operation of mechanical equipment and production safety in industrial environments.However,due to the nonlinearity and non-stationarity of collected vibration signals,single-modal methods struggle to capture fault features fully.This paper proposes a rolling bearing fault diagnosis method based on multi-modal information fusion.The method first employs the Hippopotamus Optimization Algorithm(HO)to optimize the number of modes in Variational Mode Decomposition(VMD)to achieve optimal modal decomposition performance.It combines Convolutional Neural Networks(CNN)and Gated Recurrent Units(GRU)to extract temporal features from one-dimensional time-series signals.Meanwhile,the Markovian Transition Field(MTF)is used to transform one-dimensional signals into two-dimensional images for spatial feature mining.Through visualization techniques,the effectiveness of generated images from different parameter combinations is compared to determine the optimal parameter configuration.A multi-modal network(GSTCN)is constructed by integrating Swin-Transformer and the Convolutional Block Attention Module(CBAM),where the attention module is utilized to enhance fault features.Finally,the fault features extracted from different modalities are deeply fused and fed into a fully connected layer to complete fault classification.Experimental results show that the GSTCN model achieves an average diagnostic accuracy of 99.5%across three datasets,significantly outperforming existing comparison methods.This demonstrates that the proposed model has high diagnostic precision and good generalization ability,providing an efficient and reliable solution for rolling bearing fault diagnosis.
基金supported by the National Natural Science Foundation of China(Nos.12405288,12374241,12474484,U2330401,12088101)the Natural Science Foundation of Top Talent of SZTU(No.GDRC202526)。
文摘The process of nuclear fusion in the presence of a laser field was theoretically analyzed.The analysis is applicable to most fusion reactions and different types of currently available intense lasers,from X-ray free-electron lasers to solid-state near-infrared lasers.Laser fields were shown to enhance the fusion yields,and the mechanism of this enhancement was explained.Low-frequency lasers are more efficient in enhancing fusion than high-frequency lasers.The calculation results show enhancements of fusion yields by orders of magnitude with currently available intense low-frequency laser fields.The temperature requirement for controlled nuclear fusion may be reduced with the aid of intense laser fields.
文摘Unmanned aerial vehicles(UAVs),especially quadcopters,have become indispensable in numerous industrial and scientific applications due to their flexibility,lowcost,and capability to operate in dynamic environments.This paper presents a complete design and implementation of a compact autonomous quadcopter capable of trajectory tracking,object detection,precision landing,and real-time telemetry via long-range communication protocols.The system integrates an onboard flight controller running real-time sensor fusion algorithms,a vision-based detection system on a companion single-board computer,and a telemetry unit using Long Range(LoRa)communication.Extensive flight tests were conducted to validate the system’s stability,communication range,and autonomous capabilities.Potential applications in law enforcement,agriculture,search and rescue,and environmental monitoring are also discussed.
基金the financial support from the National Key Research and Development Program of China(Grant No.2022YFB4400100)the NSFC under Grant Nos.92477102 and 62122084the open research fund of Songshan Lake Materials Laboratory 2023SLABFK09。
文摘The increasing complexity of intelligent sensing environments,driven by the growth of Internet of Things technologies,has created a strong demand for neuromorphic systems capable of real-time,low-power multisensory perception.Traditional sensory architectures,constrained by single-modal processing and centralized computing,struggle to meet the requirements of diverse and dynamic input conditions.Multisensory neuromorphic devices offer a promising solution by mimicking the distributed,event-driven processing of biological systems.Recent efforts have explored synaptic devices and material systems that respond to various input modalities,including visual,tactile,thermal,and chemical stimuli.However,challenges remain in signal conversion,encoding compatibility,and the fusion of heterogeneous inputs without loss of unisensory information.This review provides a comprehensive overview of the physical mechanisms,device behaviors,and integration strategies that underpin signal processing in neuromorphic hardware.We highlight synaptic mechanisms conducive to cross-modal interaction,analyze representative signal fusion approaches at the device level,and discuss future directions for constructing efficient,scalable,and biologically inspired multisensory neuromorphic systems.
基金supported by the National Natural Science Foundation of China(Grant Nos.62572057,62272049,U24A20331)Beijing Natural Science Foundation(Grant Nos.4232026,4242020)Academic Research Projects of Beijing Union University(Grant No.ZK10202404).
文摘Traffic sign detection is a critical component of driving systems.Single-stage network-based traffic sign detection algorithms,renowned for their fast detection speeds and high accuracy,have become the dominant approach in current practices.However,in complex and dynamic traffic scenes,particularly with smaller traffic sign objects,challenges such as missed and false detections can lead to reduced overall detection accuracy.To address this issue,this paper proposes a detection algorithm that integrates edge and shape information.Recognizing that traffic signs have specific shapes and distinct edge contours,this paper introduces an edge feature extraction branch within the backbone network,enabling adaptive fusion with features of the same hierarchical level.Additionally,a shape prior convolution module is designed to replaces the first two convolutional modules of the backbone network,aimed at enhancing the model's perception ability for specific shape objects and reducing its sensitivity to background noise.The algorithm was evaluated on the CCTSDB and TT100k datasets,and compared to YOLOv8s,the mAP50 values increased by 3.0%and 10.4%,respectively,demonstrating the effectiveness of the proposed method in improving the accuracy of traffic sign detection.
基金supported by the Defense Industrial Technology Development Program(Grant No.JCKY2022212C002)the National Natural Science Foundation of China(Grant No.52225503)+1 种基金the National Natural Science Foundation of China for Creative Research Groups(Grant No.51921003)the Fundamental Research Funds for the Central Universities(Grant No.NI2024003)。
文摘Tungsten(W),owing to its exceptional physical and chemical properties,is a promising candidate for plasma-facing materials.However,its intrinsic brittleness makes it highly susceptible to cracking during its processing,especially using the laser powder bed fusion(LPBF)process with high energy input.Alloying has been recognized as an effective strategy to deal with this challenge,yet investigations into rare earth element alloying for tungsten and corresponding LPBF production remain limited.In this study,yttrium(Y),a rare earth element,was introduced to alloy with tungsten,and W-Y alloys were fabricated via LPBF at various laser energy densities.Finite element simulations were conducted to predict the temperature field distributions of W-Y alloy under different laser energy densities,providing insight into the formation of metallurgical defects at various laser energy inputs.It revealed that at a suitable laser energy density of 500 J/mm^(3),the fabricated W-Y specimens exhibited smooth and flat melt paths without discernible internal pores or cracks,achieving densification of 99.3%.The W-Y alloy had a refined microstructure with fine columnar and equiaxial grains,with an average grain size of 15.83μm.The compressive strength and elongation after fracture of the W-Y alloy were 1531.93 MPa and 21.57%,respectively.The excellent hardness of 520 HV_(0.2)and wear resistance with a coefficient of friction(COF)of 0.47 were obtained.The enhanced mechanical performance could be attributed to grain refinement strengthening and dislocation strengthening.The study on the tensile properties of the intrinsic brittle W-Y sample was conducted for the first time,highlighting the intrinsic challenge of improving the tensile ductility of tungsten.This study not only provides a new perspective on rare-earth-alloyed tungsten but also offers a scientific reference for LPBF processing of high-performance W-Y alloy.
基金funded by the National Natural Science Foundation of China,grant number 62262045the Fundamental Research Funds for the Central Universities,grant number 2023CDJYGRH-YB11the Open Funding of SUGON Industrial Control and Security Center,grant number CUIT-SICSC-2025-03.
文摘Autonomous vehicles rely heavily on accurate and efficient scene segmentation for safe navigation and efficient operations.Traditional Bird’s Eye View(BEV)methods on semantic scene segmentation,which leverage multimodal sensor fusion,often struggle with noisy data and demand high-performance GPUs,leading to sensor misalignment and performance degradation.This paper introduces an Enhanced Channel Attention BEV(ECABEV),a novel approach designed to address the challenges under insufficient GPU memory conditions.ECABEV integrates camera and radar data through a de-noise enhanced channel attention mechanism,which utilizes global average and max pooling to effectively filter out noise while preserving discriminative features.Furthermore,an improved fusion approach is proposed to efficiently merge categorical data across modalities.To reduce computational overhead,a bilinear interpolation layer normalizationmethod is devised to ensure spatial feature fidelity.Moreover,a scalable crossentropy loss function is further designed to handle the imbalanced classes with less computational efficiency sacrifice.Extensive experiments on the nuScenes dataset demonstrate that ECABEV achieves state-of-the-art performance with an IoU of 39.961,using a lightweight ViT-B/14 backbone and lower resolution(224×224).Our approach highlights its cost-effectiveness and practical applicability,even on low-end devices.The code is publicly available at:https://github.com/YYF-CQU/ECABEV.git.
文摘The initial noise present in the depth images obtained with RGB-D sensors is a combination of hardware limitations in addition to the environmental factors,due to the limited capabilities of sensors,which also produce poor computer vision results.The common image denoising techniques tend to remove significant image details and also remove noise,provided they are based on space and frequency filtering.The updated framework presented in this paper is a novel denoising model that makes use of Boruta-driven feature selection using a Long Short-Term Memory Autoencoder(LSTMAE).The Boruta algorithm identifies the most useful depth features that are used to maximize the spatial structure integrity and reduce redundancy.An LSTMAE is then used to process these selected features and model depth pixel sequences to generate robust,noise-resistant representations.The system uses the encoder to encode the input data into a latent space that has been compressed before it is decoded to retrieve the clean image.Experiments on a benchmark data set show that the suggested technique attains a PSNR of 45 dB and an SSIM of 0.90,which is 10 dB higher than the performance of conventional convolutional autoencoders and 15 times higher than that of the wavelet-based models.Moreover,the feature selection step will decrease the input dimensionality by 40%,resulting in a 37.5%reduction in training time and a real-time inference rate of 200 FPS.Boruta-LSTMAE framework,therefore,offers a highly efficient and scalable system for depth image denoising,with a high potential to be applied to close-range 3D systems,such as robotic manipulation and gesture-based interfaces.
基金supported in part by the by Chongqing Research Program of Basic Research and Frontier Technology under Grant CSTB2025NSCQ-GPX1309.
文摘Small object detection has been a focus of attention since the emergence of deep learning-based object detection.Although classical object detection frameworks have made significant contributions to the development of object detection,there are still many issues to be resolved in detecting small objects due to the inherent complexity and diversity of real-world visual scenes.In particular,the YOLO(You Only Look Once)series of detection models,renowned for their real-time performance,have undergone numerous adaptations aimed at improving the detection of small targets.In this survey,we summarize the state-of-the-art YOLO-based small object detection methods.This review presents a systematic categorization of YOLO-based approaches for small-object detection,organized into four methodological avenues,namely attention-based feature enhancement,detection-head optimization,loss function,and multi-scale feature fusion strategies.We then examine the principal challenges addressed by each category.Finally,we analyze the performance of thesemethods on public benchmarks and,by comparing current approaches,identify limitations and outline directions for future research.
基金supported by the National Natural Science Foundation of China Grant(No.61972133)Project of Leading Talents in Science and Technology Innovation for Thousands of People Plan in Henan Province Grant(No.204200510021)the Key Research and Development Plan Special Project of Henan Province Grant(No.241111211400).
文摘At inference time,deep neural networks are susceptible to backdoor attacks,which can produce attackercontrolled outputs when inputs contain carefully crafted triggers.Existing defense methods often focus on specific attack types or incur high costs,such as data cleaning or model fine-tuning.In contrast,we argue that it is possible to achieve effective and generalizable defense without removing triggers or incurring high model-cleaning costs.Fromthe attacker’s perspective and based on characteristics of vulnerable neuron activation anomalies,we propose an Adaptive Feature Injection(AFI)method for black-box backdoor detection.AFI employs a pre-trained image encoder to extract multi-level deep features and constructs a dynamic weight fusionmechanism for precise identification and interception of poisoned samples.Specifically,we select the control samples with the largest feature differences fromthe clean dataset via feature-space analysis,and generate blended sample pairs with the test sample using dynamic linear interpolation.The detection statistic is computed by measuring the divergence G(x)in model output responses.We systematically evaluate the effectiveness of AFI against representative backdoor attacks,including BadNets,Blend,WaNet,and IAB,on three benchmark datasets:MNIST,CIFAR-10,and ImageNet.Experimental results show that AFI can effectively detect poisoned samples,achieving average detection rates of 95.20%,94.15%,and 86.49%on these datasets,respectively.Compared with existing methods,AFI demonstrates strong cross-domain generalization ability and robustness to unknown attacks.
基金funded by the Hainan Province Science and Technology Special Fund under Grant ZDYF2024GXJS292.
文摘Deep learning has made significant progress in the field of oriented object detection for remote sensing images.However,existing methods still face challenges when dealing with difficult tasks such as multi-scale targets,complex backgrounds,and small objects in remote sensing.Maintaining model lightweight to address resource constraints in remote sensing scenarios while improving task completion for remote sensing tasks remains a research hotspot.Therefore,we propose an enhanced multi-scale feature extraction lightweight network EM-YOLO based on the YOLOv8s architecture,specifically optimized for the characteristics of large target scale variations,diverse orientations,and numerous small objects in remote sensing images.Our innovations lie in two main aspects:First,a dynamic snake convolution(DSC)is introduced into the backbone network to enhance the model’s feature extraction capability for oriented targets.Second,an innovative focusing-diffusion module is designed in the feature fusion neck to effectively integrate multi-scale feature information.Finally,we introduce Layer-Adaptive Sparsity for magnitude-based Pruning(LASP)method to perform lightweight network pruning to better complete tasks in resource-constrained scenarios.Experimental results on the lightweight platform Orin demonstrate that the proposed method significantly outperforms the original YOLOv8s model in oriented remote sensing object detection tasks,and achieves comparable or superior performance to state-of-the-art methods on three authoritative remote sensing datasets(DOTA v1.0,DOTA v1.5,and HRSC2016).
