论文阅读笔记:“Coherence Constrained Graph LSTM for Group Activity Recognition”
sixwalter Lv6

Coherence Constrained Graph LSTM for Group Activity Recognition (TPAMI 2022)

主要思想:寻找群体体行为中的相关行为

STCC和GCC约束

  • STCC:如果一个个体行为在时间域上相互关联,并且在空间域上与其他个体行为具有一致性,我们称该种行为为相关行为
  • GCC:如果一个特定的行为与整体群体行为一致性更高,那么它对群体行为的贡献也越高
image-20230417095353737

网络框架

image-20230417100222815

核心模块

spatial interactions among persons are dependent on the temporal motions of individuals

CCG-LSTM

时空场景关联推理

image-20230417103246814 image-20230417112848658 image-20230417113033797 image-20230417124514121

temporal confidence gate

image-20230417140102656 image-20230417112107287
  • 如果相差越小,值越接近1
  • 加入这个temporal confidence gate后的新公式:
image-20230417124355537

spatial context confidence gate

捕获空间域的个体的“relative”运动

image-20230417140407896 image-20230417140424319 image-20230417140440639
  • 如果相差越小,值越接近1
  • 加入这个spatial confidence gate后的新公式:
image-20230417140728207

AGG-LSTM

全局场景关联推理

image-20230417103201208
  • Post title:论文阅读笔记:“Coherence Constrained Graph LSTM for Group Activity Recognition”
  • Post author:sixwalter
  • Create time:2023-04-17 00:00:00
  • Post link:https://coelien.github.io/2023/04/17/paper-reading/paper_reading_063/
  • Copyright Notice:All articles in this blog are licensed under BY-NC-SA unless stating additionally.
 Comments