【论文笔记】2020CVPR - PREDICT & CLUSTER: Unsupervised Skeleton Based Action Recognition

本文转载自查看原文 2020-10-16 19:48 543 Action Recognition

任务

action recognition

对骨架节点坐标进行视角不变预处理

We align the action sequences by implementing a view-invariant transformation which transforms keypoints coordinates from original coordinate system into a view-invariant coordinate

system。

基本流程

先训练出一个encoder-decoder，这个encoder是双向GRU，decoder是单向GRU。把encoder得到的特征叫做high dimensional feature。如下图：

再把high-dimensional feature送进另一个encoder-decoder，这里的encoder和decoder都是全连接层，激活函数是tanh，如下图所示。把这个encoder得到的特征叫做low dimensional feature。用这个low dimensional feature做KNN聚类。

数据集

North-Western UCLA (NW-UCLA) dataset

UWA3D Multiview Activity II (UWA3D) dataset

NTU RGB+D dataset

免责声明！

本站转载的文章为个人学习借鉴使用，本站对版权不负任何法律责任。如果侵犯了您的隐私权益，请联系本站邮箱yoyou2525@163.com删除。

猜您在找 【论文翻译】：2020-CVPR-PREDICT & CLUSTER: Unsupervised Skeleton Based Action Recognition 【论文笔记】Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition 【论文笔记，action recognition,动作识别】论文笔记 | A Closer Look at Spatiotemporal Convolutions for Action Recognition Skeleton-Based Action Recognition with Directed Graph Neural Network 论文笔记：Unsupervised Domain Adaptation by Backpropagation Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition (ST-GCN) 论文笔记【MDNet//CVPR2016】 Two-Stream Adaptive Graph Convolutional Network for Skeleton-Based Action Recognition GHM论文笔记（CVPR2019）