SOTAVerified

3D Multi-Person Pose Estimation (root-relative)

This task aims to solve root-relative 3D multi-person pose estimation (person-centric coordinate system). No ground truth human bounding box and human root joint coordinates are used during testing stage.

( Image credit: RootNet )

Papers

Showing 1118 of 18 papers

TitleStatusHype
Towards 3D Human Pose Estimation in the Wild: a Weakly-supervised ApproachCode0
Permutation-Invariant Relational Network for Multi-person 3D Pose Estimation0
PI-Net: Pose Interacting Network for Multi-Person Monocular 3D Pose Estimation0
Towards Robust and Smooth 3D Multi-Person Pose Estimation from Monocular Videos in the Wild0
Dynamic Graph Reasoning for Multi-person 3D Pose Estimation0
Deep Monocular 3D Human Pose Estimation via Cascaded Dimension-Lifting0
HMOR: Hierarchical Multi-Person Ordinal Relations for Monocular Multi-Person 3D Pose Estimation0
Explicit Occlusion Reasoning for Multi-person 3D Human Pose Estimation0
Show:102550
← PrevPage 2 of 2Next →

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1LCR-NetMPJPE146Unverified
2Mehta et al.MPJPE132Unverified
3Dual network3DPCK89.6Unverified
4TDBU_Net3DPCK89.6Unverified
5GnTCN3DPCK87.5Unverified
6PIRN3DPCK85.8Unverified
7Temporal Smoothing3DPCK85.3Unverified
8GR-M3D3DPCK84.6Unverified
9POTR-3D3DPCK83.7Unverified
10HDNet3DPCK83.7Unverified