SOTAVerified

3D Multi-Person Pose Estimation

This task aims to solve root-relative 3D multi-person pose estimation. No human bounding box and root joint coordinate groundtruth are used in testing time.

( Image credit: RootNet )

Papers

Showing 110 of 62 papers

TitleStatusHype
Multi-person Physics-based Pose Estimation for Combat Sports0
RapidPoseTriangulation: Multi-view Multi-person Whole-body Human Pose Triangulation in a MillisecondCode0
Multi-HMR: Multi-Person Whole-Body Human Mesh Recovery in a Single ShotCode3
Instance-aware Contrastive Learning for Occluded Human Mesh ReconstructionCode1
Geometry-Biased Transformer for Robust Multi-View 3D Human Pose Reconstruction0
Towards Robust and Smooth 3D Multi-Person Pose Estimation from Monocular Videos in the Wild0
Three Recipes for Better 3D Pseudo-GTs of 3D Human Mesh Estimation in the WildCode1
A Unified Multi-view Multi-person Tracking Framework0
Weakly Supervised 3D Multi-person Pose Estimation for Large-scale Scenes based on Monocular Camera and Single LiDAR0
Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and RefinementCode1
Show:102550
← PrevPage 1 of 7Next →

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1Multi-Person 3D Pose and Shape Estimation via Inverse Kinematics and Refinement3DPCK89.9Unverified
2Multi-HMR3DPCK89.5Unverified
3Unsupervised Cross-Modal Alignment3DPCK78.4Unverified
4MUG3DPCK76.27Unverified
53DCrowdNet3DPCK76.2Unverified
6SelecSLS3DPCK75.8Unverified
7HG-RCNN3DPCK74.2Unverified
8BMP3DPCK73.83Unverified
93DCrowdNet (HigherHRNet)3DPCK72.7Unverified