ID 原文 译文
24715 行人重识别旨在跨监控设备下检索出特定的行人目标 。为捕捉行人图像的多粒度特征进而提高识别精度，基于 OSNet 基准网络提出一种多粒度特征融合网络(Multi-granularity Feature Fusion Network for Person Re-Identi-fication, MFN)进行端对端的学习。 For the purpose of capturing the multi-granularity features and improving the recognition accuracy, amulti-granularity feature fusion network for person re-identification (MFN) is proposed based on the omist-scale network (OSNet).
24716 MFN 由全局分支、特征擦除分支和局部分支组成，其中特征擦除分支由双通道注意力擦除模型构成，此模型包含通道注意力擦除模块(Channel Attention-based Dropout Moudle, CDM)和空间注意力擦除模块(Spatial Attention-based Dropout Moudle, SDM)。 The MFN network is composed of a global branch, a feature dropout branch and a local branch. The feature dropout branch consists of a dual-channel attention dropout model, which includes a channel attention-based dropout moudle (CDM) and a Spatial attention-based dropout moudle (SDM).
24717 CDM对通道的注意力强度排序并擦除低注意力通道，SDM在空间维度上以一定概率擦除最具有判别力的特征，两者通过并联方式相互作用，提高模型的识别能力。 CDM sorts the attention intensity and dropouts low attention channels, and SDM dropouts the most discriminative features with a certain probability in the spatial dimension.
24718 全局分支采用特征金字塔结构提取多尺度特征，局部分支将特征均匀切块后级联成一个单一特征，提取关键局部信息。 The global branch uses the feature pyramid structure to extract multi-scale features, and the local branch employs a uniform partition strategy to produce local features which are cascaded into a single one for key local information extraction.
24719 大量实验结果表明了本文方法的有效性，在 Market1501、DukeMTMC-reID 和 CUHK03-Labeled(Detected)数据集上，mAP/Rank-1 分别达到了90.1%/95.8%、81.8%/91.4%和80.7%/82.3%(78.7%/81.6%)，大幅优于其他现有方法。 Experiments on the large scale datasets show the effectiveness of MFN. On the Market1501, DukeMTMC-reID and CUHK03-Labeled (Detected) datasets, mAP/Rank-1 of MFN reaches 90.1%/95.8%, 81.8%/91.4% and 80.7%/82.3% (78.7%/81.6%), which is superior to other existing methods.
24720 为了提高分析信号的信噪比，本文提出了一种基于变分模态分解的变步长归一化最小均方自适应滤波降噪方法。 To improve the SNR of received signals, a normalized minimum mean square adaptive filtering denoising method based on variational mode decomposition using variable step-size was proposed.
24721 该方法对原信号进行变分模态分解并区分信号分量和噪声分量，再对噪声分量进行间隙阈值降噪处理并将其作为参考信号输入自适应滤波器，通过自适应算法迭代处理得到降噪后的信号分量，并通过重构算法得到最终降噪后的信号。 The proposed algorithm decomposed the original signal into several components labelled as noise or signal component. Then an interval threshold denoising method was exploited to denoise the noise component before being inputted into an adaptive filter as a reference signal. All the rest signal components were used to reconstruct the final denoised signal after being denoised by iterative adaptive filters.
24722 本文还在变分模态分解的基础上使用小波阈值降噪和间隙阈值降噪方法按不同方案进行降噪处理并得到最佳算法。 In addition, an optimal algorithm based on variational mode decomposition using wavelet threshold denoising and interval threshold denoising methods was exploited.
24723 实验结果表明，本文所提自适应滤波降噪方法的降噪效果比阈值降噪最佳方法效果更好。 Experimental results show that the proposed adaptive filtering denoising method outperforms the optimal algorithm using threshold denoising.
24724 针对低压配电网箱表关系存在人工核查成本高、异常案例少、难以实现异常规律捕获的问题，采用极端不平衡分类学习方法实现低压异常箱表关系识别的泛化应用推广。 Due to high labor cost and few abnormal cases of power box-table relations inspection, which is difficult to obtain the law. The extreme unbalanced classification learning method was used to capture the generalization.