As an integrated application of modern information technologies and artificial intelligence,Prognostic and Health Management(PHM)is important for machine health monitoring.Prediction of tool wear is one of the symboli...As an integrated application of modern information technologies and artificial intelligence,Prognostic and Health Management(PHM)is important for machine health monitoring.Prediction of tool wear is one of the symbolic applications of PHM technology in modern manufacturing systems and industry.In this paper,a multi-scale Convolutional Gated Recurrent Unit network(MCGRU)is proposed to address raw sensory data for tool wear prediction.At the bottom of MCGRU,six parallel and independent branches with different kernel sizes are designed to form a multi-scale convolutional neural network,which augments the adaptability to features of different time scales.These features of different scales extracted from raw data are then fed into a Deep Gated Recurrent Unit network to capture long-term dependencies and learn significant representations.At the top of the MCGRU,a fully connected layer and a regression layer are built for cutting tool wear prediction.Two case studies are performed to verify the capability and effectiveness of the proposed MCGRU network and results show that MCGRU outperforms several state-of-the-art baseline models.展开更多
To perform landslide susceptibility prediction(LSP),it is important to select appropriate mapping unit and landslide-related conditioning factors.The efficient and automatic multi-scale segmentation(MSS)method propose...To perform landslide susceptibility prediction(LSP),it is important to select appropriate mapping unit and landslide-related conditioning factors.The efficient and automatic multi-scale segmentation(MSS)method proposed by the authors promotes the application of slope units.However,LSP modeling based on these slope units has not been performed.Moreover,the heterogeneity of conditioning factors in slope units is neglected,leading to incomplete input variables of LSP modeling.In this study,the slope units extracted by the MSS method are used to construct LSP modeling,and the heterogeneity of conditioning factors is represented by the internal variations of conditioning factors within slope unit using the descriptive statistics features of mean,standard deviation and range.Thus,slope units-based machine learning models considering internal variations of conditioning factors(variant slope-machine learning)are proposed.The Chongyi County is selected as the case study and is divided into 53,055 slope units.Fifteen original slope unit-based conditioning factors are expanded to 38 slope unit-based conditioning factors through considering their internal variations.Random forest(RF)and multi-layer perceptron(MLP)machine learning models are used to construct variant Slope-RF and Slope-MLP models.Meanwhile,the Slope-RF and Slope-MLP models without considering the internal variations of conditioning factors,and conventional grid units-based machine learning(Grid-RF and MLP)models are built for comparisons through the LSP performance assessments.Results show that the variant Slopemachine learning models have higher LSP performances than Slope-machine learning models;LSP results of variant Slope-machine learning models have stronger directivity and practical application than Grid-machine learning models.It is concluded that slope units extracted by MSS method can be appropriate for LSP modeling,and the heterogeneity of conditioning factors within slope units can more comprehensively reflect the relationships between conditioning factors and landslides.The research results have important reference significance for land use and landslide prevention.展开更多
This presentation predicts the elastic properties of three-dimensional(3D)orthogonal woven composite(3DOWC)by finite element analysis based on micro/meso repeated unit cell(RUC)models.First,the properties of fiber yar...This presentation predicts the elastic properties of three-dimensional(3D)orthogonal woven composite(3DOWC)by finite element analysis based on micro/meso repeated unit cell(RUC)models.First,the properties of fiber yarn are obtained by analysis on a micro-scale RUC model assuming fibers in a hexagonal distribution pattern in the polymer matrix.Then a full thickness meso-scale RUC model including weft yarns,warp yarns,Z-yarns and pure resin zones is established and full stiffness matrix of the 3DOWC including the in-plane and flexural constants are predicted.For thick 3DOWC with large number of weft,warp layers,an alternative analysis method is proposed in which an inner meso-RUC and a surface meso-RUC are established,respectively.Then the properties of 3DOWC are deduced based on laminate theory and properties of the inner and surface layers.The predicted results by the above two alternative methods are in good experimental agreement.展开更多
Recently,speech enhancement methods based on Generative Adversarial Networks have achieved good performance in time-domain noisy signals.However,the training of Generative Adversarial Networks has such problems as con...Recently,speech enhancement methods based on Generative Adversarial Networks have achieved good performance in time-domain noisy signals.However,the training of Generative Adversarial Networks has such problems as convergence difficulty,model collapse,etc.In this work,an end-to-end speech enhancement model based on Wasserstein Generative Adversarial Networks is proposed,and some improvements have been made in order to get faster convergence speed and better generated speech quality.Specifically,in the generator coding part,each convolution layer adopts different convolution kernel sizes to conduct convolution operations for obtaining speech coding information from multiple scales;a gated linear unit is introduced to alleviate the vanishing gradient problem with the increase of network depth;the gradient penalty of the discriminator is replaced with spectral normalization to accelerate the convergence rate of themodel;a hybrid penalty termcomposed of L1 regularization and a scale-invariant signal-to-distortion ratio is introduced into the loss function of the generator to improve the quality of generated speech.The experimental results on both TIMIT corpus and Tibetan corpus show that the proposed model improves the speech quality significantly and accelerates the convergence speed of the model.展开更多
针对菌落图像中小菌落易漏检的问题,提出了一种基于INC4-YOLO(you only look once)的计数方法,实现精准的菌落计数。采用带残差结构的Inception模块(Inception module with residual connection,IncRes)替换YOLOv5骨干网络中的Bottlenec...针对菌落图像中小菌落易漏检的问题,提出了一种基于INC4-YOLO(you only look once)的计数方法,实现精准的菌落计数。采用带残差结构的Inception模块(Inception module with residual connection,IncRes)替换YOLOv5骨干网络中的Bottleneck模块,以增强图像特征提取能力。从网络的浅层特征中引出一个小目标检测头,以增强算法在训练过程中对小菌落的注意力。分别在标注微生物自动识别数据集(annotated germs for automated recognition,AGAR)和真实菌落计数场景下对INC4-YOLO进行计数性能测试。实验结果表明,在AGAR测试集中,提出的算法在小菌落的平均百分比绝对值计数误差(mean absolute percentage error,MAPE)比其他先进目标检测算法降低了2%;真实菌落计数场景下,INC4-YOLO的MAPE相比YOLOv5降低了7%,表明该算法可帮助菌落计数设备实现精准计数。展开更多
Accurate prediction of the state-of-charge(SOC)of battery energy storage system(BESS)is critical for its safety and lifespan in electric vehicles.To overcome the imbalance of existing methods between multi-scale featu...Accurate prediction of the state-of-charge(SOC)of battery energy storage system(BESS)is critical for its safety and lifespan in electric vehicles.To overcome the imbalance of existing methods between multi-scale feature fusion and global feature extraction,this paper introduces a novel multi-scale fusion(MSF)model based on gated recurrent unit(GRU),which is specifically designed for complex multi-step SOC prediction in practical BESSs.Pearson correlation analysis is first employed to identify SOC-related parameters.These parameters are then input into a multi-layer GRU for point-wise feature extraction.Concurrently,the parameters undergo patching before entering a dual-stage multi-layer GRU,thus enabling the model to capture nuanced information across varying time intervals.Ultimately,by means of adaptive weight fusion and a fully connected network,multi-step SOC predictions are rendered.Following extensive validation over multiple days,it is illustrated that the proposed model achieves an absolute error of less than 1.5%in real-time SOC prediction.展开更多
新一代视频编码标准高效视频编码(High Efficiency Video Coding,HEVC)中编码单元(Coding Unit,CU)大小不同的特性使得编码效率得到显著提升,但同时带来了极高的计算复杂度.为了去除CU划分中多余的计算从而降低编码复杂度,本文提出了一...新一代视频编码标准高效视频编码(High Efficiency Video Coding,HEVC)中编码单元(Coding Unit,CU)大小不同的特性使得编码效率得到显著提升,但同时带来了极高的计算复杂度.为了去除CU划分中多余的计算从而降低编码复杂度,本文提出了一种利用深度学习的编码单元快速划分算法.首先使用原始视频亮度块及编码信息建立了一个HEVC中CU划分的数据库,用于接下来本文深度学习神经网络的训练.然后,为了更好地贴合编码单元划分的层级结构,本文提出了一种基于Inception模块的神经网络结构,使之内嵌于HEVC编码框架中对编码单元的划分进行提前预测,有效地去除了All Intra配置下中冗余的CU划分计算.实验结果表明,本文提出的算法与HEVC官方测试模型(HM16.12)相比,编码时间平均降低了61.31%,而BD-BR与BD-PSNR仅为1.86%和-0.13dB.展开更多
The location of the inception point of the air entrainment directly affects the energy dissipation ratio, the cavitation damage control, and the training wall height designs for a stepped spillway and a stilling basin...The location of the inception point of the air entrainment directly affects the energy dissipation ratio, the cavitation damage control, and the training wall height designs for a stepped spillway and a stilling basin. In this paper, the boundary layer theory of plates is used to predict the location of the inception point of the air entrainment over the stepped spillways by assuming the steps on the spillways as a kind of roughness. An empirical formula is presented based on the physical model experiments, with the maximum error less than 1% except at one point where the error is 1.6%, as compared to the experimental data. Meanwhile, it is shown that the location of the inception point of the air entrainment for the stepped spillway is much nearer to the top of the spillway than that for a smooth spillways, which explains why the high ratio of the energy dissipation is provided for the stepped spillway.展开更多
A supercomputer with 1.0 Petaflops peak performance in single precision, designed and established by Institute of Process Engineering, Chinese Academy of Sciences, is introduced in this brief communication. A designin...A supercomputer with 1.0 Petaflops peak performance in single precision, designed and established by Institute of Process Engineering, Chinese Academy of Sciences, is introduced in this brief communication. A designing philosophy utilizing the similarity between hardware, software and the problems to be solved is embodied, based on the multi-scale method and discrete simulation approaches developed at Institute of Process Engineering （IPE） and implemented in a graphic processing unit （GPU）-based hybrid computing mode. The preliminary applications of this machine in areas of multi-phase flow, molecular dynamics and so on are reported, demonstrating the supercomputer as a paradigm of green computation in new architecture.展开更多
With the increasing intensive and large-scale development of the sika deer breeding industry,it is crucial to assess the health status of the sika deer by monitoring their behaviours.A machine vision-based method for ...With the increasing intensive and large-scale development of the sika deer breeding industry,it is crucial to assess the health status of the sika deer by monitoring their behaviours.A machine vision-based method for the behaviour recognition of sika deer is proposed in this paper.Google Inception Net(GoogLeNet)is used to optimise the model in this paper.First,the number of layers and size of the model were reduced.Then,the 5×5 convolution was changed to two 3×3 convolutions,which reduced the parameters and increased the nonlinearity of the model.A 5×5 convolution kernel was used to replace the original convolution for extracting coarse-grained features and improving the model’s extraction ability.A multi-scale module was added to the model to enhance the multi-faceted feature extraction capability of the model.Simultaneously,the Squeeze-and-Excitation Networks(SE-Net)module was included to increase the channel’s attention and improve the model’s accuracy.The dataset’s images were rotated to reduce overfitting.For image rotation,the angle wasmultiplied by 30°to obtain the dataset enhanced by rotation operations of 30°,60°,90°,120°and 150°.The experimental results showed that the recognition rate of this model in the behaviour of sika deer was 98.92%.Therefore,the model presented in this paper can be applied to the behaviour recognition of sika deer.The results will play an essential role in promoting animal behaviour recognition technology and animal health monitoring management.展开更多
Nowadays,the COVID-19 virus disease is spreading rampantly.There are some testing tools and kits available for diagnosing the virus,but it is in a lim-ited count.To diagnose the presence of disease from radiological i...Nowadays,the COVID-19 virus disease is spreading rampantly.There are some testing tools and kits available for diagnosing the virus,but it is in a lim-ited count.To diagnose the presence of disease from radiological images,auto-mated COVID-19 diagnosis techniques are needed.The enhancement of AI(Artificial Intelligence)has been focused in previous research,which uses X-ray images for detecting COVID-19.The most common symptoms of COVID-19 are fever,dry cough and sore throat.These symptoms may lead to an increase in the rigorous type of pneumonia with a severe barrier.Since medical imaging is not suggested recently in Canada for critical COVID-19 diagnosis,computer-aided systems are implemented for the early identification of COVID-19,which aids in noticing the disease progression and thus decreases the death rate.Here,a deep learning-based automated method for the extraction of features and classi-fication is enhanced for the detection of COVID-19 from the images of computer tomography(CT).The suggested method functions on the basis of three main pro-cesses:data preprocessing,the extraction of features and classification.This approach integrates the union of deep features with the help of Inception 14 and VGG-16 models.At last,a classifier of Multi-scale Improved ResNet(MSI-ResNet)is developed to detect and classify the CT images into unique labels of class.With the support of available open-source COVID-CT datasets that consists of 760 CT pictures,the investigational validation of the suggested method is estimated.The experimental results reveal that the proposed approach offers greater performance with high specificity,accuracy and sensitivity.展开更多
由于背景环境复杂,检测物体易受部分遮挡、天气以及光线变化等因素的影响,传统目标检测方法存在提取特征难、检测准确率低、检测耗时长等缺陷.为了改善传统目标检测方法存在的缺陷,实现快速准确的目标检测,提出了一种基于快速区域卷积...由于背景环境复杂,检测物体易受部分遮挡、天气以及光线变化等因素的影响,传统目标检测方法存在提取特征难、检测准确率低、检测耗时长等缺陷.为了改善传统目标检测方法存在的缺陷,实现快速准确的目标检测,提出了一种基于快速区域卷积神经网络(faster regions with convolutional neural network,Faster-RCNN)算法的轻量化改进方法,即针对算法Inception-V2特征提取网络进行轻量化改进,并以带泄露线性整流(leaky rectified linear unit,Leaky ReLU)作为激活函数,解决使用线性整流(rectified linear unit,ReLU)激活函数存在的神经元输入为负数时输出为0的问题.基于上述改进方法,选择沙滩废弃物的检测为案例以验证方法的有效性,并且结合不同特征提取网络在检测沙滩废弃物时的表现,对比了SSD(single shot multibox detector)与Faster-RCNN算法.实验结果表明:所提改进算法在实际检测中有较好的综合性能,且相比原算法Faster-RCNN_Inception-V2,轻量化改进后的Inception-V2特征提取网络卷积计算量减少51.8%,模型训练耗时缩短了9.1%,检测耗时减少了10.9%,各类别AP的平均值(mean average precision,mAP)增加了1.02%,可见所提的改进方法能够有效提高目标检测的准确率,减少检测耗时,并在沙滩废弃物检测上得到成功应用,为海滨城市的沙滩清理维护提供了技术支持与保障.展开更多
Deep learning for time series sequence individual data instance classification can revolutionize computer assisted navigation by providing surgeons with accurate, real-time instrument locality through automatic instru...Deep learning for time series sequence individual data instance classification can revolutionize computer assisted navigation by providing surgeons with accurate, real-time instrument locality through automatic instrument localization. This paper presents an evaluation of Deep Learning models to perform individual data instance classification of time series data. The models explored include convolution and recurrent networks, as well as state-of-the-art residual and inception architectures. The time series data used to evaluate the models consists of depth and force measurements from a drill. Four recurrent neural network models using long short-term memory and gated recurrent units, known as baseline models, and four models using 1D convolution with ResNet and Inception architectures, known as advanced models, were evaluated by determining the data instance membership of the four classes. The four classes represent four distinct regions in a bone traversed by the drill bit during a surgical procedure. First, the time series data is preprocessed, identifying the four classes or regions of the bone. Next, the paper presents a discussion of the network architecture and modifications of both the basic and advanced deep learning models, followed by the training process and hyperparameters tuning. The performance of the models was evaluated using the precision and recall performance parameters. Out of the eight models evaluated, the recurrent neural network with gated recurrent units has the best performance. The paper also demonstrates the importance of the feature depth over the feature force in classifying the data instances, followed by the effects of the imbalanced dataset on the performance of the models.展开更多
基金Supported in part by Natural Science Foundation of China(Grant Nos.51835009,51705398)Shaanxi Province 2020 Natural Science Basic Research Plan(Grant No.2020JQ-042)Aeronautical Science Foundation(Grant No.2019ZB070001).
文摘As an integrated application of modern information technologies and artificial intelligence,Prognostic and Health Management(PHM)is important for machine health monitoring.Prediction of tool wear is one of the symbolic applications of PHM technology in modern manufacturing systems and industry.In this paper,a multi-scale Convolutional Gated Recurrent Unit network(MCGRU)is proposed to address raw sensory data for tool wear prediction.At the bottom of MCGRU,six parallel and independent branches with different kernel sizes are designed to form a multi-scale convolutional neural network,which augments the adaptability to features of different time scales.These features of different scales extracted from raw data are then fed into a Deep Gated Recurrent Unit network to capture long-term dependencies and learn significant representations.At the top of the MCGRU,a fully connected layer and a regression layer are built for cutting tool wear prediction.Two case studies are performed to verify the capability and effectiveness of the proposed MCGRU network and results show that MCGRU outperforms several state-of-the-art baseline models.
基金funded by the Natural Science Foundation of China(Grant Nos.41807285,41972280 and 52179103).
文摘To perform landslide susceptibility prediction(LSP),it is important to select appropriate mapping unit and landslide-related conditioning factors.The efficient and automatic multi-scale segmentation(MSS)method proposed by the authors promotes the application of slope units.However,LSP modeling based on these slope units has not been performed.Moreover,the heterogeneity of conditioning factors in slope units is neglected,leading to incomplete input variables of LSP modeling.In this study,the slope units extracted by the MSS method are used to construct LSP modeling,and the heterogeneity of conditioning factors is represented by the internal variations of conditioning factors within slope unit using the descriptive statistics features of mean,standard deviation and range.Thus,slope units-based machine learning models considering internal variations of conditioning factors(variant slope-machine learning)are proposed.The Chongyi County is selected as the case study and is divided into 53,055 slope units.Fifteen original slope unit-based conditioning factors are expanded to 38 slope unit-based conditioning factors through considering their internal variations.Random forest(RF)and multi-layer perceptron(MLP)machine learning models are used to construct variant Slope-RF and Slope-MLP models.Meanwhile,the Slope-RF and Slope-MLP models without considering the internal variations of conditioning factors,and conventional grid units-based machine learning(Grid-RF and MLP)models are built for comparisons through the LSP performance assessments.Results show that the variant Slopemachine learning models have higher LSP performances than Slope-machine learning models;LSP results of variant Slope-machine learning models have stronger directivity and practical application than Grid-machine learning models.It is concluded that slope units extracted by MSS method can be appropriate for LSP modeling,and the heterogeneity of conditioning factors within slope units can more comprehensively reflect the relationships between conditioning factors and landslides.The research results have important reference significance for land use and landslide prevention.
基金BASTRI Subtopic Research about Digital Sampler Technology of Body Structure Performance Study Based on Big Data Calculation Model,China(No.MIIT Civil aircraft special purpose MJ-2017-F-20)
文摘This presentation predicts the elastic properties of three-dimensional(3D)orthogonal woven composite(3DOWC)by finite element analysis based on micro/meso repeated unit cell(RUC)models.First,the properties of fiber yarn are obtained by analysis on a micro-scale RUC model assuming fibers in a hexagonal distribution pattern in the polymer matrix.Then a full thickness meso-scale RUC model including weft yarns,warp yarns,Z-yarns and pure resin zones is established and full stiffness matrix of the 3DOWC including the in-plane and flexural constants are predicted.For thick 3DOWC with large number of weft,warp layers,an alternative analysis method is proposed in which an inner meso-RUC and a surface meso-RUC are established,respectively.Then the properties of 3DOWC are deduced based on laminate theory and properties of the inner and surface layers.The predicted results by the above two alternative methods are in good experimental agreement.
基金supported by the National Science Foundation under Grant No.62066039.
文摘Recently,speech enhancement methods based on Generative Adversarial Networks have achieved good performance in time-domain noisy signals.However,the training of Generative Adversarial Networks has such problems as convergence difficulty,model collapse,etc.In this work,an end-to-end speech enhancement model based on Wasserstein Generative Adversarial Networks is proposed,and some improvements have been made in order to get faster convergence speed and better generated speech quality.Specifically,in the generator coding part,each convolution layer adopts different convolution kernel sizes to conduct convolution operations for obtaining speech coding information from multiple scales;a gated linear unit is introduced to alleviate the vanishing gradient problem with the increase of network depth;the gradient penalty of the discriminator is replaced with spectral normalization to accelerate the convergence rate of themodel;a hybrid penalty termcomposed of L1 regularization and a scale-invariant signal-to-distortion ratio is introduced into the loss function of the generator to improve the quality of generated speech.The experimental results on both TIMIT corpus and Tibetan corpus show that the proposed model improves the speech quality significantly and accelerates the convergence speed of the model.
文摘针对菌落图像中小菌落易漏检的问题,提出了一种基于INC4-YOLO(you only look once)的计数方法,实现精准的菌落计数。采用带残差结构的Inception模块(Inception module with residual connection,IncRes)替换YOLOv5骨干网络中的Bottleneck模块,以增强图像特征提取能力。从网络的浅层特征中引出一个小目标检测头,以增强算法在训练过程中对小菌落的注意力。分别在标注微生物自动识别数据集(annotated germs for automated recognition,AGAR)和真实菌落计数场景下对INC4-YOLO进行计数性能测试。实验结果表明,在AGAR测试集中,提出的算法在小菌落的平均百分比绝对值计数误差(mean absolute percentage error,MAPE)比其他先进目标检测算法降低了2%;真实菌落计数场景下,INC4-YOLO的MAPE相比YOLOv5降低了7%,表明该算法可帮助菌落计数设备实现精准计数。
基金supported in part by the National Natural Science Foundation of China(No.62172036).
文摘Accurate prediction of the state-of-charge(SOC)of battery energy storage system(BESS)is critical for its safety and lifespan in electric vehicles.To overcome the imbalance of existing methods between multi-scale feature fusion and global feature extraction,this paper introduces a novel multi-scale fusion(MSF)model based on gated recurrent unit(GRU),which is specifically designed for complex multi-step SOC prediction in practical BESSs.Pearson correlation analysis is first employed to identify SOC-related parameters.These parameters are then input into a multi-layer GRU for point-wise feature extraction.Concurrently,the parameters undergo patching before entering a dual-stage multi-layer GRU,thus enabling the model to capture nuanced information across varying time intervals.Ultimately,by means of adaptive weight fusion and a fully connected network,multi-step SOC predictions are rendered.Following extensive validation over multiple days,it is illustrated that the proposed model achieves an absolute error of less than 1.5%in real-time SOC prediction.
基金Project supported by the National Natural Science Foundation of China(Grant No.51179114)
文摘The location of the inception point of the air entrainment directly affects the energy dissipation ratio, the cavitation damage control, and the training wall height designs for a stepped spillway and a stilling basin. In this paper, the boundary layer theory of plates is used to predict the location of the inception point of the air entrainment over the stepped spillways by assuming the steps on the spillways as a kind of roughness. An empirical formula is presented based on the physical model experiments, with the maximum error less than 1% except at one point where the error is 1.6%, as compared to the experimental data. Meanwhile, it is shown that the location of the inception point of the air entrainment for the stepped spillway is much nearer to the top of the spillway than that for a smooth spillways, which explains why the high ratio of the energy dissipation is provided for the stepped spillway.
文摘A supercomputer with 1.0 Petaflops peak performance in single precision, designed and established by Institute of Process Engineering, Chinese Academy of Sciences, is introduced in this brief communication. A designing philosophy utilizing the similarity between hardware, software and the problems to be solved is embodied, based on the multi-scale method and discrete simulation approaches developed at Institute of Process Engineering （IPE） and implemented in a graphic processing unit （GPU）-based hybrid computing mode. The preliminary applications of this machine in areas of multi-phase flow, molecular dynamics and so on are reported, demonstrating the supercomputer as a paradigm of green computation in new architecture.
基金This research is supported by the Science and Technology Department of Jilin Province[20210202128NC http://kjt.jl.gov.cn]The People’s Republic of China Ministry of Science and Technology[2018YFF0213606-03 http://www.most.gov.cn]+1 种基金Jilin Province Development and Reform Commission[2019C021 http://jldrc.jl.gov.cn]the Science and Technology Bureau of Changchun City[21ZGN27 http://kjj.changchun.gov.cn].
文摘With the increasing intensive and large-scale development of the sika deer breeding industry,it is crucial to assess the health status of the sika deer by monitoring their behaviours.A machine vision-based method for the behaviour recognition of sika deer is proposed in this paper.Google Inception Net(GoogLeNet)is used to optimise the model in this paper.First,the number of layers and size of the model were reduced.Then,the 5×5 convolution was changed to two 3×3 convolutions,which reduced the parameters and increased the nonlinearity of the model.A 5×5 convolution kernel was used to replace the original convolution for extracting coarse-grained features and improving the model’s extraction ability.A multi-scale module was added to the model to enhance the multi-faceted feature extraction capability of the model.Simultaneously,the Squeeze-and-Excitation Networks(SE-Net)module was included to increase the channel’s attention and improve the model’s accuracy.The dataset’s images were rotated to reduce overfitting.For image rotation,the angle wasmultiplied by 30°to obtain the dataset enhanced by rotation operations of 30°,60°,90°,120°and 150°.The experimental results showed that the recognition rate of this model in the behaviour of sika deer was 98.92%.Therefore,the model presented in this paper can be applied to the behaviour recognition of sika deer.The results will play an essential role in promoting animal behaviour recognition technology and animal health monitoring management.
基金Supporting this research through Taif University Researchers Supporting Project number(TURSP-2020/231),Taif University,Taif,Saudi Arabia.
文摘Nowadays,the COVID-19 virus disease is spreading rampantly.There are some testing tools and kits available for diagnosing the virus,but it is in a lim-ited count.To diagnose the presence of disease from radiological images,auto-mated COVID-19 diagnosis techniques are needed.The enhancement of AI(Artificial Intelligence)has been focused in previous research,which uses X-ray images for detecting COVID-19.The most common symptoms of COVID-19 are fever,dry cough and sore throat.These symptoms may lead to an increase in the rigorous type of pneumonia with a severe barrier.Since medical imaging is not suggested recently in Canada for critical COVID-19 diagnosis,computer-aided systems are implemented for the early identification of COVID-19,which aids in noticing the disease progression and thus decreases the death rate.Here,a deep learning-based automated method for the extraction of features and classi-fication is enhanced for the detection of COVID-19 from the images of computer tomography(CT).The suggested method functions on the basis of three main pro-cesses:data preprocessing,the extraction of features and classification.This approach integrates the union of deep features with the help of Inception 14 and VGG-16 models.At last,a classifier of Multi-scale Improved ResNet(MSI-ResNet)is developed to detect and classify the CT images into unique labels of class.With the support of available open-source COVID-CT datasets that consists of 760 CT pictures,the investigational validation of the suggested method is estimated.The experimental results reveal that the proposed approach offers greater performance with high specificity,accuracy and sensitivity.
文摘由于背景环境复杂,检测物体易受部分遮挡、天气以及光线变化等因素的影响,传统目标检测方法存在提取特征难、检测准确率低、检测耗时长等缺陷.为了改善传统目标检测方法存在的缺陷,实现快速准确的目标检测,提出了一种基于快速区域卷积神经网络(faster regions with convolutional neural network,Faster-RCNN)算法的轻量化改进方法,即针对算法Inception-V2特征提取网络进行轻量化改进,并以带泄露线性整流(leaky rectified linear unit,Leaky ReLU)作为激活函数,解决使用线性整流(rectified linear unit,ReLU)激活函数存在的神经元输入为负数时输出为0的问题.基于上述改进方法,选择沙滩废弃物的检测为案例以验证方法的有效性,并且结合不同特征提取网络在检测沙滩废弃物时的表现,对比了SSD(single shot multibox detector)与Faster-RCNN算法.实验结果表明:所提改进算法在实际检测中有较好的综合性能,且相比原算法Faster-RCNN_Inception-V2,轻量化改进后的Inception-V2特征提取网络卷积计算量减少51.8%,模型训练耗时缩短了9.1%,检测耗时减少了10.9%,各类别AP的平均值(mean average precision,mAP)增加了1.02%,可见所提的改进方法能够有效提高目标检测的准确率,减少检测耗时,并在沙滩废弃物检测上得到成功应用,为海滨城市的沙滩清理维护提供了技术支持与保障.
文摘Deep learning for time series sequence individual data instance classification can revolutionize computer assisted navigation by providing surgeons with accurate, real-time instrument locality through automatic instrument localization. This paper presents an evaluation of Deep Learning models to perform individual data instance classification of time series data. The models explored include convolution and recurrent networks, as well as state-of-the-art residual and inception architectures. The time series data used to evaluate the models consists of depth and force measurements from a drill. Four recurrent neural network models using long short-term memory and gated recurrent units, known as baseline models, and four models using 1D convolution with ResNet and Inception architectures, known as advanced models, were evaluated by determining the data instance membership of the four classes. The four classes represent four distinct regions in a bone traversed by the drill bit during a surgical procedure. First, the time series data is preprocessed, identifying the four classes or regions of the bone. Next, the paper presents a discussion of the network architecture and modifications of both the basic and advanced deep learning models, followed by the training process and hyperparameters tuning. The performance of the models was evaluated using the precision and recall performance parameters. Out of the eight models evaluated, the recurrent neural network with gated recurrent units has the best performance. The paper also demonstrates the importance of the feature depth over the feature force in classifying the data instances, followed by the effects of the imbalanced dataset on the performance of the models.
