论文阅读笔记:“Spatiotemporal Multiplier Networks for Video Action Recognition”
Spatiotemporal Multiplier Networks for Video Action Recognition
主要思想
- to solve the combination of two streams:
multiplicative motion gating(乘法运动门控)
如图2d,其公式为:
在反向传播过程中,运动和表观流的输入被显示地牵涉,作为其梯度上的门控机制。它使得模型有能力去学习时空特征关联
- to solve long-term input
inject temporal filters(注入时序过滤器)
对应该变换的初始化,作者令其不会改变原始特征(即恒等映射),思想其实就是进行一个3D卷积,只在通道上和时序上进行了信息处理(不涉及空间维度信息变换)。
- Post title:论文阅读笔记:“Spatiotemporal Multiplier Networks for Video Action Recognition”
- Post author:sixwalter
- Create time:2023-08-05 11:14:26
- Post link:https://coelien.github.io/2023/08/05/paper-reading/paper_reading_057/
- Copyright Notice:All articles in this blog are licensed under BY-NC-SA unless stating additionally.
Comments