针对独居老人摔倒问题,构建一种基于双流卷积神经网络(TwoStream CNN)的实时跌倒检测模型。将提取人物轮廓的RGB单帧作为输入的空间流,将连续多帧运动历史图(motion history image,MHI)作为输入的时间流融合,在一个特定维度的全连接层...针对独居老人摔倒问题,构建一种基于双流卷积神经网络(TwoStream CNN)的实时跌倒检测模型。将提取人物轮廓的RGB单帧作为输入的空间流,将连续多帧运动历史图(motion history image,MHI)作为输入的时间流融合,在一个特定维度的全连接层将两个网络的同shape张量Concatenation相连后添加到新的层,在公开的摔倒数据集上进行实验和定量分析。实验结果表明,采用人物轮廓RGB-MHI的双流卷积网络在非摔倒和摔倒的区分检测中准确率达到了98.12%,改进的融合方式相比较其它方法有提高,时间流输入MHI满足实时性要求。展开更多
A signal detection scheme was proposed for two-way relaying networks（TWRNs） using distributed differential space-time coding（DDSTC） under imperfect synchronization. Unlike most existing work perfect with synchroni...A signal detection scheme was proposed for two-way relaying networks（TWRNs） using distributed differential space-time coding（DDSTC） under imperfect synchronization. Unlike most existing work perfect with synchronization assumed, a relative delay between the signals transmitted from both sources to the relay was considered. Since perfect channel state information（CSI） is difficult to be acquired in fast fading, the scenarios and computation complexity will be increased especially when there appear multiple relays, CSI is assumed unavailable at all nodes. Therefore, the article proposes a differential signal detection scheme based on estimating and cancelling the imperfect synchronization component in the received signal at the two source nodes, followed by a least square（LS） decoder. Simulations, using the Nakagami-m fading channel due to its versatile statistical distribution property, show that the proposed scheme for both source nodes are effective in suppressing the inter-symbol interference（ISI） caused by imperfect synchronization while neither the source nodes nor the relay nodes have any knowledge of CSI.展开更多
文摘针对独居老人摔倒问题,构建一种基于双流卷积神经网络(TwoStream CNN)的实时跌倒检测模型。将提取人物轮廓的RGB单帧作为输入的空间流,将连续多帧运动历史图(motion history image,MHI)作为输入的时间流融合,在一个特定维度的全连接层将两个网络的同shape张量Concatenation相连后添加到新的层,在公开的摔倒数据集上进行实验和定量分析。实验结果表明,采用人物轮廓RGB-MHI的双流卷积网络在非摔倒和摔倒的区分检测中准确率达到了98.12%,改进的融合方式相比较其它方法有提高,时间流输入MHI满足实时性要求。
基金supported by the Shanghai Natural Science Foundation (14ZR1415100)key laboratory of specialty fiber optics and optical access networks, Shanghai University, (SKLSFO2012-14)+1 种基金funding of key laboratory of wireless sensor network & communication,shanghai institute of microsystem and information technologyfunding of Shanghai Education Committee, Chinese academy of sciences and Shanghai science committee (12511503303, 14511105602 ,14511105902)
文摘A signal detection scheme was proposed for two-way relaying networks（TWRNs） using distributed differential space-time coding（DDSTC） under imperfect synchronization. Unlike most existing work perfect with synchronization assumed, a relative delay between the signals transmitted from both sources to the relay was considered. Since perfect channel state information（CSI） is difficult to be acquired in fast fading, the scenarios and computation complexity will be increased especially when there appear multiple relays, CSI is assumed unavailable at all nodes. Therefore, the article proposes a differential signal detection scheme based on estimating and cancelling the imperfect synchronization component in the received signal at the two source nodes, followed by a least square（LS） decoder. Simulations, using the Nakagami-m fading channel due to its versatile statistical distribution property, show that the proposed scheme for both source nodes are effective in suppressing the inter-symbol interference（ISI） caused by imperfect synchronization while neither the source nodes nor the relay nodes have any knowledge of CSI.
