Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Absolute Human Pose Estimation with Depth Prediction Network

About

The common approach to 3D human pose estimation is predicting the body joint coordinates relative to the hip. This works well for a single person but is insufficient in the case of multiple interacting people. Methods predicting absolute coordinates first estimate a root-relative pose then calculate the translation via a secondary optimization task. We propose a neural network that predicts joints in a camera centered coordinate system instead of a root-relative one. Unlike previous methods, our network works in a single step without any post-processing. Our network beats previous methods on the MuPoTS-3D dataset and achieves state-of-the-art results.

M\'arton V\'eges, Andr\'as L\H{o}rincz• 2019

Related benchmarks

TaskDatasetResultRank
Multi-person 3D Pose EstimationMuPoTS-3D (test)--
41
3D Multi-person Pose EstimationMuPoTS-3D
3D PCK Score75.4
21
3D Human Pose EstimationMuPoTS-3D unnormalized skeletons
R-MPJPE120
4
Showing 3 of 3 rows

Other info

Code

Follow for update