TSwinPose: Enhanced monocular 3D human pose estimation with JointFlow

Open Webpage

Muyu Li, Henan Hu, Jingjing Xiong, Xudong Zhao, Hong Yan

Published: 01 Jan 2024, Last Modified: 13 Nov 2024Expert Syst. Appl. 2024EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•TSwinPose, a novel multi-scale spatio-temporal monocular 3D pose estimation method.•JointFlow, a plug-in extension which boosts 2D-to-3D lifting human pose estimation.•Thorough analysis shows SOTA results and confirms the performance boost of JointFlow.