BriefGPT.xyz
Sep, 2019
三维人体姿态估计的跨视角融合
Cross View Fusion for 3D Human Pose Estimation
HTML
PDF
Haibo Qiu, Chunyu Wang, Jingdong Wang, Naiyan Wang, Wenjun Zeng
TL;DR
该研究提出了一种基于多视角几何先验的方法，通过逐步改进三维姿态的准确性来从多视角图像中恢复绝对三维人体姿态。
Abstract
We present an approach to recover absolute
3d human poses
from
multi-view images
by incorporating multi-view
geometric priors
in our model
→