...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>IEEE sensors journal >Pyramidal Graph Convolutional Network for Skeleton-Based Human Action Recognition
【24h】

Pyramidal Graph Convolutional Network for Skeleton-Based Human Action Recognition

机译：基于骨架的人类行动识别的金字塔图形图卷积网络

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

The emergence of low-cost depth sensors opens up new potentials for skeleton-based human action recognition. The recent methods for this task have made significant progress by incorporating graph convolution. However, they (1) have limitations in modeling the complex and variable temporal dynamics, and (2) cannot maximize the complementarity of the spatial and temporal features. Besides, (3) the loss function of these methods has an inherent weakness in optimizing the intraclass compactness. To this end, we propose a pyramidal graph convolutional network (PY-GCN) in this paper. Specifically, (1) an effective yet efficient single-oriented pyramidal convolution is proposed. It involves multiple kernels with varying sizes and depths that are capable of capturing different levels of the temporal dynamics at multiple scales. (2) A pseudo-two-stream structure for the basic block of the network is proposed to comprehensively aggregate discriminative cross-spatiotemporal features. Moreover, (3) a pairwise Gaussian loss together with the cross-entropy loss is introduced to the model, which can focus on both intraclass compactness and interclass separability. Our PY-GCN achieves state-of-the-art performance on three challenging large-scale datasets.
机译：低成本深度传感器的出现为基于骨架的人类行动识别开辟了新的潜力。最近对此任务的方法通过纳入图形卷积来取得了重大进展。但是，它们（1）对复杂和可变时间动态进行建模，并且（2）不能最大化空间和时间特征的互补性。此外，（3）这些方法的损耗功能在优化薄层的紧凑性方面具有固有的弱点。为此，我们提出了本文的金字塔图形卷积网络（PY-GCN）。具体地，（1）提出了一种有效但有效的单型金字塔型卷积。它涉及多个内核，其具有不同尺寸和深度，其能够在多个尺度处捕获不同级别的时间动态。 （2）提出了一种用于网络基本块的伪两流结构，以综合聚集鉴别性的交叉时空特征。此外，（3）将双向高斯损耗与跨熵损耗一起引入模型，这可以专注于跨越紧凑性和嵌段可分离性。我们的PY-GCN在三个具有挑战性的大型数据集上实现了最先进的性能。

著录项

  • 来源
    《IEEE sensors journal》 |2021年第14期|16183-16191|共9页
  • 作者

    Li Fanjia; Zhu Aichun; Liu Zhongyu; Huo Yu; Xu Yonggang; Hua Gang;

  • 作者单位

    China Univ Min & Technol Sch Informat & Control Engn Xuzhou 221116 Jiangsu Peoples R China|Jiangsu Prov Xuzhou Technician Inst Xuzhou 221151 Jiangsu Peoples R China;

    Nanjing Tech Univ Sch Comp Sci & Technol Nanjing 211800 Peoples R China;

    China Univ Min & Technol Sch Informat & Control Engn Xuzhou 221116 Jiangsu Peoples R China|China Univ Min & Technol IOT Percept Mine Res Ctr Xuzhou 221116 Jiangsu Peoples R China;

    China Univ Min & Technol Sch Informat & Control Engn Xuzhou 221116 Jiangsu Peoples R China|China Univ Min & Technol IOT Percept Mine Res Ctr Xuzhou 221116 Jiangsu Peoples R China;

    China Univ Min & Technol Sch Informat & Control Engn Xuzhou 221116 Jiangsu Peoples R China;

    China Univ Min & Technol Sch Informat & Control Engn Xuzhou 221116 Jiangsu Peoples R China;

  • 收录信息
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类
  • 关键词

    Skeleton-based; action recognition; pyramidal convolution; pseudo-two-stream;

    机译：基于骨架的;动作识别;金字塔卷积;伪两条;

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号