A Discriminative Channel Diversification Network for Image Classification.
A Discriminative Channel Diversification Network for Image Classification.
复制标题
DOI：
10.1016/j.patrec.2021.12.004
复制
发表时间：
2022-01
影响因子：
5.1
通讯作者：
Wang, Guanghui
中科院分区：
文献类型：
--
作者：
Patel, Krushi;Wang, Guanghui
卷积神经网络中的通道注意力机制已被证明在各种计算机视觉任务中有效。然而，性能的提高伴随着额外的模型复杂度和计算成本。在本文中，我们提出了一个轻量级和有效的注意模块，称为渠道多样化块，以提高全球范围内建立渠道关系在全球范围内。与其他通道注意机制不同，所提出的模块通过在考虑通道激活的同时更多地关注空间可区分的通道来关注最具区别性的特征。与其他注意力模型将注意力模块插入中间层不同，该模型将注意力模块嵌入到骨干网络的末端，易于实现。在CIFAR-10、SVHN和Tiny-ImageNet数据集上进行的大量实验表明，该模块平均将基线网络的性能提高了3%。
Channel attention mechanisms in convolutional neural networks have been proven to be effective in various computer vision tasks. However, the performance improvement comes with additional model complexity and computation cost. In this paper, we propose a light-weight and effective attention module, called channel diversification block, to enhance the global context by establishing the channel relationship at the global level. Unlike other channel attention mechanisms, the proposed module focuses on the most discriminative features by giving more attention to the spatially distinguishable channels while taking account of the channel activation. Different from other attention models that plugin the module in between several intermediate layers, the proposed module is embedded at the end of the backbone networks, making it easy to implement. Extensive experiments on CIFAR-10, SVHN, and Tiny-ImageNet datasets demonstrate that the proposed module improves the performance of the baseline networks by a margin of 3% on average.
登录
查看更多内容
影响因子：
6
作者：
He, Lei;Lu, Jiwen;Zhou, Jie
通讯作者：
Zhou, Jie
影响因子：
5.1
作者：
Guan, Qingji;Huang, Yaping;Yang, Yi
通讯作者：
Yang, Yi
影响因子：
8
作者：
Cen, Feng;Zhao, Xiaoyu;Wang, Guanghui
通讯作者：
Wang, Guanghui
影响因子：
8
作者：
Ma, Wenchi;Wu, Yuanwei;Wang, Guanghui
通讯作者：
Wang, Guanghui
影响因子：
22.7
作者：
Krizhevsky, Alex;Sutskever, Ilya;Hinton, Geoffrey E.
通讯作者：
Hinton, Geoffrey E.