BriefGPT.xyz
Aug, 2019
深度视频3D人体姿势估计的轨迹空间分解
Trajectory Space Factorization for Deep Video-Based 3D Human Pose Estimation
HTML
PDF
Jiahao Lin, Gim Hee Lee
TL;DR
该论文提出了一种使用矩阵分解来进行序列3D人体姿势估计的深度学习框架,通过对轨迹系数矩阵进行深度回归操作,可以为输入序列中的每个帧输出3D姿态估计,取得了多个基准数据集上最先进的表现。
Abstract
Existing
deep learning
approaches on
3d human pose estimation
for videos are either based on Recurrent or Convolutional Neural Networks (RNNs or CNNs). However, RNN-based frameworks can only tackle sequences with
→