A Divide-and-conquer Solution to 3D Human Motion Estimation from Raw MoCap Data
Jilin Tang, Lincheng Li, Jie Hou, Haoran Xin, Xin Feng Yu · 2023
Marker-based optical motion capture (MoCap) aims to estimate 3D human motions from a sequence of input raw markers. In this paper, we propose a divide-and-conquer strategy based MoCap solving network that accurately retrieves 3D human skeleton motions from raw marker sequences in real-time. Our core idea is to decompose the task of direct estimation of global human motion from all markers into first solving sub-motions of local parts and then aggregating sub-motions into a global one to achieve accurate motion estimation.