As the use of deepfake facial videos proliferate,the associated threats to social security and integrity cannot be overstated.Effective methods for detecting forged facial videos are thus urgently needed.While many de...As the use of deepfake facial videos proliferate,the associated threats to social security and integrity cannot be overstated.Effective methods for detecting forged facial videos are thus urgently needed.While many deep learning-based facial forgery detection approaches show promise,they often fail to delve deeply into the complex relationships between image features and forgery indicators,limiting their effectiveness to specific forgery techniques.To address this challenge,we propose a dual-branch collaborative deepfake detection network.The network processes video frame images as input,where a specialized noise extraction module initially extracts the noise feature maps.Subsequently,the original facial images and corresponding noise maps are directed into two parallel feature extraction branches to concurrently learn texture and noise forgery clues.An attention mechanism is employed between the two branches to facilitate mutual guidance and enhancement of texture and noise features across four different scales.This dual-modal feature integration enhances sensitivity to forgery artifacts and boosts generalization ability across various forgery techniques.Features from both branches are then effectively combined and processed through a multi-layer perception layer to distinguish between real and forged video.Experimental results on benchmark deepfake detection datasets demonstrate that our approach outperforms existing state-of-the-art methods in terms of detection performance,accuracy,and generalization ability.展开更多
Chronic kidney disease(CKD) is a widespread renal disease throughout the world. Once it develops to the advanced stage, serious complications and high risk of death will follow. Hence, early screening is crucial for t...Chronic kidney disease(CKD) is a widespread renal disease throughout the world. Once it develops to the advanced stage, serious complications and high risk of death will follow. Hence, early screening is crucial for the treatment of CKD. Since ultrasonography has no side effects and enables radiologists to dynamically observe the morphology and pathological features of the kidney, it is commonly used for kidney examination. In this study,we propose a novel convolutional neural network(CNN) framework named the texture branch network to screen CKD based on ultrasound images. This introduces a texture branch into a typical CNN to extract and optimize texture features. The model can automatically generate texture features and deep features from input images, and use the fused information as the basis of classification. Furthermore, we train the base part of the network by means of transfer learning, and conduct experiments on a dataset with 226 ultrasound images. Experimental results demonstrate the effectiveness of the proposed approach, achieving an accuracy of 96.01% and a sensitivity of 99.44%.展开更多
Asymmetric tree-like branched networks are explored by geometric algorithms. Based on the network, an analysis of the thermal conductivity is presented. The relationship between effective thermal conductivity and geom...Asymmetric tree-like branched networks are explored by geometric algorithms. Based on the network, an analysis of the thermal conductivity is presented. The relationship between effective thermal conductivity and geometric structures is obtained by using the thermal-electrical analogy technique. In all studied cases, a clear behaviour is observed, where angle （δ,θ） among parent branching extended lines, branches and parameter of the geometric structures have stronger effects on the effective thermal conductivity. When the angle δ is fixed, the optical diameter ratio β＋ is dependent on angle θ. Moreover, γand m are not related to β＊. The longer the branch is, the smaller the effective thermal conductivity will be. It is also found that when the angle θ〈δ2, the higher the iteration m is, the lower the thermal conductivity will be and it tends to zero, otherwise, it is bigger than zero. When the diameter ratio β1 〈 0.707 and angle δ is bigger, the optimal k of the perfect ratio increases with the increase of the angle δ; when β1 〉 0.707, the optimal k decreases. In addition, the effective thermal conductivity is always less than that of single channel material. The present results also show that the effective thermal conductivity of the asymmetric tree-like branched networks does not obey Murray＇s law.展开更多
The matrix D describing relations of the loops to the nodes in the graph and also the setsof branches based on the independent loops and their matrix Q are defined.The theorem in whichthe product of the loop-node matr...The matrix D describing relations of the loops to the nodes in the graph and also the setsof branches based on the independent loops and their matrix Q are defined.The theorem in whichthe product of the loop-node matrix D multiplied by the incidence matrix A<sub>a</sub> is equal to matrix Qis put forward and proved.The admittance matrix Y<sub>lc</sub> of the sets of the branches is defined and it isassumed that the vector V<sub>lc</sub> of voltage of the sets of branches to be a calculative quantity.The equa-tion of the sets of branches is derived and the analysis method of the sets of branches based on theindependent loops in the electric network is presented.展开更多
Convolutional neural networks （CNNs） have been applied in state-of-the-art visual tracking tasks to represent the target. However, most existing algorithms treat visual tracking as an object-specific task. Therefore...Convolutional neural networks （CNNs） have been applied in state-of-the-art visual tracking tasks to represent the target. However, most existing algorithms treat visual tracking as an object-specific task. Therefore, the model needs to be retrained for different test video sequences. We propose a branch-activated multi-domain convolutional neural network （BAMDCNN）. In contrast to most existing trackers based on CNNs which require frequent online training, BAMDCNN only needs offine training and online fine-tuning. Specifically, BAMDCNN exploits category-specific features that are more robust against variations. To allow for learning category-specific information, we introduce a group algorithm and a branch activation method. Experimental results on challenging benchmark show that the proposed algorithm outperforms other state-of-the-art methods. What＇s more, compared with CNN based trackers, BAMDCNN increases tracking speed.展开更多
目的 针对目前三维人体姿态估计方法未能有效处理时间序列冗余,难以捕获人体关节上微小变化的问题,提出一种融合多关节特征的单目视觉三维人体姿态估计网络。方法 在关节运动特征提取模块中,采用多分支操作提取关节在时间维度上的运动特...目的 针对目前三维人体姿态估计方法未能有效处理时间序列冗余,难以捕获人体关节上微小变化的问题,提出一种融合多关节特征的单目视觉三维人体姿态估计网络。方法 在关节运动特征提取模块中,采用多分支操作提取关节在时间维度上的运动特征,并将不同特征融合形成具有高度表达力的特征表示。关节特征融合模块整合了不同关节组和中间帧的全局信息,通过矩阵内积的方式表达不同关节组在高纬度空间的相对位置及相互联系,得到中间3D姿态的初估值。关节约束模块引入中间帧的2D关节点空间位置关系作为隐式约束,与中间帧3D姿态初估值融合,减少不合理的姿态输出,提高最终3D姿态估计的准确性。结果 实验结果表明,与MHFormer方法相比,本文方法在Human3.6M数据集上的平均关节位置误差(mean per joint position error,MPJPE)结果为29.0 mm,误差降低4.9%,对于复杂动作,如SittingDown和WalkDog,误差降低了7.7%和8.2%。在MPI-INF-3DHP数据集上,MPJPE指标降低36.2%,曲线下面积(area under the curve,AUC)指标提升12.9%,正确关节点百分比(percentage of correct keypoints,PCK)指标提升3%。实验结果体现出在面对复杂动作问题时,网络利用各分支提取了不同的关节时序运动特征,将不同关节组的位置信息进行融合交互,结合当前帧的关节姿态信息加以约束,取得更高的精度。在HumanEva数据集上的实验结果表明了本文方法适用不同数据集,消融实验进一步验证了各个模块的有效性。结论 本文网络有效地融合了人体多关节特征,可以更好地提高单目视觉三维人体姿态估计的准确性,且具备较高的泛化性。展开更多
为了提高利用监控和数据采集(supervisory control and data acquisition,SCADA)多变量长时间序列预测齿轮箱油温的精度,解决不同风电机组因处不同运行环境导致的数据分布不一致的问题,提出了一种基于多分支时间序列预测与迁移学习相结...为了提高利用监控和数据采集(supervisory control and data acquisition,SCADA)多变量长时间序列预测齿轮箱油温的精度,解决不同风电机组因处不同运行环境导致的数据分布不一致的问题,提出了一种基于多分支时间序列预测与迁移学习相结合的齿轮箱状态监测方法。首先,利用极致梯度提升(extreme gradient boosting,XGBoost)算法筛选输入参数组成原始序列,对其进行分解得到季节与趋势序列。其次,提出季节、趋势序列特征提取模块获取季节及趋势特征的序列,将其与经过Informer模型处理后的特征序列进行融合后输入进多层感知机映射成最终的预测值,以构建提出的多分支时间序列预测网络(multi-branch time series prediction network,MBFN)。最后,利用迁移学习并结合一分类向量支持机(one-class support vector machine,OCSVM)模型及滑动窗口构建齿轮箱的健康指数,完成齿轮箱状态监测。实验结果表明,所提出模型的MBFN显著提高了油温预测精度,优于常规时间序列预测模型,所使用的迁移策略能以较少数据适应不同数据的分布,进而实现对齿轮箱的状态监测,并且所提出的模型可以提前18.9 d发出齿轮箱故障预警。展开更多
基金funded by the Ministry of Public Security Science and Technology Program Project(No.2023LL35)the Key Laboratory of Smart Policing and National Security Risk Governance,Sichuan Province(No.ZHZZZD2302).
文摘As the use of deepfake facial videos proliferate,the associated threats to social security and integrity cannot be overstated.Effective methods for detecting forged facial videos are thus urgently needed.While many deep learning-based facial forgery detection approaches show promise,they often fail to delve deeply into the complex relationships between image features and forgery indicators,limiting their effectiveness to specific forgery techniques.To address this challenge,we propose a dual-branch collaborative deepfake detection network.The network processes video frame images as input,where a specialized noise extraction module initially extracts the noise feature maps.Subsequently,the original facial images and corresponding noise maps are directed into two parallel feature extraction branches to concurrently learn texture and noise forgery clues.An attention mechanism is employed between the two branches to facilitate mutual guidance and enhancement of texture and noise features across four different scales.This dual-modal feature integration enhances sensitivity to forgery artifacts and boosts generalization ability across various forgery techniques.Features from both branches are then effectively combined and processed through a multi-layer perception layer to distinguish between real and forged video.Experimental results on benchmark deepfake detection datasets demonstrate that our approach outperforms existing state-of-the-art methods in terms of detection performance,accuracy,and generalization ability.
基金the Zhejiang Provincial Natural Science Foundation of China (No. LY18F020034)the Zhejiang Provincial Medical Health Science and Technology Project+5 种基金China(No. 2014KYB320)the National Natural Science Foundation of China (Nos. 61801428 and 61672543)the Zhejiang University Education FoundationChina (Nos. K18-511120-004 and K17-511120-017)the Major Scientific Project of Zhejiang LabChina (No. 2018DG0ZX01)。
文摘Chronic kidney disease(CKD) is a widespread renal disease throughout the world. Once it develops to the advanced stage, serious complications and high risk of death will follow. Hence, early screening is crucial for the treatment of CKD. Since ultrasonography has no side effects and enables radiologists to dynamically observe the morphology and pathological features of the kidney, it is commonly used for kidney examination. In this study,we propose a novel convolutional neural network(CNN) framework named the texture branch network to screen CKD based on ultrasound images. This introduces a texture branch into a typical CNN to extract and optimize texture features. The model can automatically generate texture features and deep features from input images, and use the fused information as the basis of classification. Furthermore, we train the base part of the network by means of transfer learning, and conduct experiments on a dataset with 226 ultrasound images. Experimental results demonstrate the effectiveness of the proposed approach, achieving an accuracy of 96.01% and a sensitivity of 99.44%.
基金Project supported by the State Key Development Program for Basic Research of China (Grant No 2006CB708612)the National Natural Science Foundation of China (Grant No 10572130)the Natural Science Foundation of Zhejiang Province, China (Grant No Y607425)
文摘Asymmetric tree-like branched networks are explored by geometric algorithms. Based on the network, an analysis of the thermal conductivity is presented. The relationship between effective thermal conductivity and geometric structures is obtained by using the thermal-electrical analogy technique. In all studied cases, a clear behaviour is observed, where angle （δ,θ） among parent branching extended lines, branches and parameter of the geometric structures have stronger effects on the effective thermal conductivity. When the angle δ is fixed, the optical diameter ratio β＋ is dependent on angle θ. Moreover, γand m are not related to β＊. The longer the branch is, the smaller the effective thermal conductivity will be. It is also found that when the angle θ〈δ2, the higher the iteration m is, the lower the thermal conductivity will be and it tends to zero, otherwise, it is bigger than zero. When the diameter ratio β1 〈 0.707 and angle δ is bigger, the optimal k of the perfect ratio increases with the increase of the angle δ; when β1 〉 0.707, the optimal k decreases. In addition, the effective thermal conductivity is always less than that of single channel material. The present results also show that the effective thermal conductivity of the asymmetric tree-like branched networks does not obey Murray＇s law.
文摘The matrix D describing relations of the loops to the nodes in the graph and also the setsof branches based on the independent loops and their matrix Q are defined.The theorem in whichthe product of the loop-node matrix D multiplied by the incidence matrix A<sub>a</sub> is equal to matrix Qis put forward and proved.The admittance matrix Y<sub>lc</sub> of the sets of the branches is defined and it isassumed that the vector V<sub>lc</sub> of voltage of the sets of branches to be a calculative quantity.The equa-tion of the sets of branches is derived and the analysis method of the sets of branches based on theindependent loops in the electric network is presented.
基金the Innovation Action Plan Foundation of Shanghai(No.16511101200)
文摘Convolutional neural networks （CNNs） have been applied in state-of-the-art visual tracking tasks to represent the target. However, most existing algorithms treat visual tracking as an object-specific task. Therefore, the model needs to be retrained for different test video sequences. We propose a branch-activated multi-domain convolutional neural network （BAMDCNN）. In contrast to most existing trackers based on CNNs which require frequent online training, BAMDCNN only needs offine training and online fine-tuning. Specifically, BAMDCNN exploits category-specific features that are more robust against variations. To allow for learning category-specific information, we introduce a group algorithm and a branch activation method. Experimental results on challenging benchmark show that the proposed algorithm outperforms other state-of-the-art methods. What＇s more, compared with CNN based trackers, BAMDCNN increases tracking speed.
文摘目的 针对目前三维人体姿态估计方法未能有效处理时间序列冗余,难以捕获人体关节上微小变化的问题,提出一种融合多关节特征的单目视觉三维人体姿态估计网络。方法 在关节运动特征提取模块中,采用多分支操作提取关节在时间维度上的运动特征,并将不同特征融合形成具有高度表达力的特征表示。关节特征融合模块整合了不同关节组和中间帧的全局信息,通过矩阵内积的方式表达不同关节组在高纬度空间的相对位置及相互联系,得到中间3D姿态的初估值。关节约束模块引入中间帧的2D关节点空间位置关系作为隐式约束,与中间帧3D姿态初估值融合,减少不合理的姿态输出,提高最终3D姿态估计的准确性。结果 实验结果表明,与MHFormer方法相比,本文方法在Human3.6M数据集上的平均关节位置误差(mean per joint position error,MPJPE)结果为29.0 mm,误差降低4.9%,对于复杂动作,如SittingDown和WalkDog,误差降低了7.7%和8.2%。在MPI-INF-3DHP数据集上,MPJPE指标降低36.2%,曲线下面积(area under the curve,AUC)指标提升12.9%,正确关节点百分比(percentage of correct keypoints,PCK)指标提升3%。实验结果体现出在面对复杂动作问题时,网络利用各分支提取了不同的关节时序运动特征,将不同关节组的位置信息进行融合交互,结合当前帧的关节姿态信息加以约束,取得更高的精度。在HumanEva数据集上的实验结果表明了本文方法适用不同数据集,消融实验进一步验证了各个模块的有效性。结论 本文网络有效地融合了人体多关节特征,可以更好地提高单目视觉三维人体姿态估计的准确性,且具备较高的泛化性。
文摘为了提高利用监控和数据采集(supervisory control and data acquisition,SCADA)多变量长时间序列预测齿轮箱油温的精度,解决不同风电机组因处不同运行环境导致的数据分布不一致的问题,提出了一种基于多分支时间序列预测与迁移学习相结合的齿轮箱状态监测方法。首先,利用极致梯度提升(extreme gradient boosting,XGBoost)算法筛选输入参数组成原始序列,对其进行分解得到季节与趋势序列。其次,提出季节、趋势序列特征提取模块获取季节及趋势特征的序列,将其与经过Informer模型处理后的特征序列进行融合后输入进多层感知机映射成最终的预测值,以构建提出的多分支时间序列预测网络(multi-branch time series prediction network,MBFN)。最后,利用迁移学习并结合一分类向量支持机(one-class support vector machine,OCSVM)模型及滑动窗口构建齿轮箱的健康指数,完成齿轮箱状态监测。实验结果表明,所提出模型的MBFN显著提高了油温预测精度,优于常规时间序列预测模型,所使用的迁移策略能以较少数据适应不同数据的分布,进而实现对齿轮箱的状态监测,并且所提出的模型可以提前18.9 d发出齿轮箱故障预警。
