Co-occurrence Feature Learning for Skeleton based Action Recognition using Regularized Deep LSTM Networks

  • 类型:arxiv
  • 标识:1603.07772
  • 链接:https://arxiv.org/abs/1603.07772
  • 主题:database
  • 主分类:multimodal
  • 形态:application
  • 被引:930
  • 被引来源:Semantic Scholar
  • S2被引:930
  • OpenAlex被引:291
  • 影响力被引:91
  • TLDR:This work takes the skeleton as the input at each time slot and introduces a novel regularization scheme to learn the co-occurrence features of skeleton joints, and proposes a new dropout algorithm which simultaneously operates on the gates, cells, and output responses of the LSTM neurons.
  • OpenAlex ID:W2307035320
  • OpenAlex DOI:10.48550/arxiv.1603.07772
  • DOI:10.48550/arxiv.1603.07772
  • DOI来源:OpenAlex
  • 开放获取:green
  • 开放获取链接:https://arxiv.org/pdf/1603.07772
  • OpenAlex更新:2026-08-19
  • 待LLM分类:否
  • 成熟度:research
  • 场景:action-recognition、skeleton、LSTM
  • 标题中文:使用正则化深度 LSTM 网络进行基于骨骼动作识别的共现特征学习
  • TLDR中文:本文在每个时间步以骨骼作为输入,引入一种新的正则化方案来学习骨骼关节的共现特征,并提出一种同时作用于 LSTM 神经元门、单元和输出响应的新型 dropout 算法。
  • 来源文件
  • [OpenAlex discover]
  • [S2 enrich]