BriefGPT.xyz
Jul, 2017
基于深度马尔可夫模型的人体姿态预测
Human Pose Forecasting via Deep Markov Models
HTML
PDF
Sam Toyer, Anoop Cherian, Tengda Han, Stephen Gould
TL;DR
本研究提出了一种基于深度马尔科夫模型的变分自编码器生成姿势的框架,探讨姿势预测的新方法,并基于一个床具拼装视频数据集进行验证。
Abstract
human pose forecasting
is an important problem in computer vision with applications to human-robot interaction, visual surveillance, and autonomous driving. Usually, forecasting algorithms use
3d skeleton sequences
→