Self-Attention Network for Skeleton-based Human Action Recognition

文章目录 摘要 介绍 文章模型概述： Self-Attention Network Muti-head attention: Approach Raw Position and Motion Data Position Data: Motion Data Encoder Non-Linear Encoder CNN Based Encoder 1.卷积第一层： 2.卷积第二层： 3.permute

>>阅读原文<<