← 4D Vision

Human-Centric 4D Modeling

SMPL-based mesh recovery and tracking, egocentric motion capture, appearance-rich human avatars, and human interaction modeling — from monocular video to HOI, HSI, and HHI.

⌘K

SMPL-based Human Mesh Recovery & Tracking 35+ papers

Model / MethodFull TitleVenueYear
SMPLA skinned multi-person linear model TOG2015
SMPLifyKeep it SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image ECCV2016
HMREnd-to-end Recovery of Human Shape and Pose CVPR2018
SMPL-XExpressive Body Capture: 3D Hands, Face, and Body from a Single Image CVPR2019
SPINLearning to Reconstruct 3D Human Pose via Model-fitting ICCV2019
GraphCMRConvolutional Mesh Regression for Single-Image Human Shape Reconstruction CVPR2019
VIBEVideo Inference for Human Body Pose and Shape Estimation CVPR2020
HybrIKHybrid Analytical-Neural Inverse Kinematics for Real-time 3D Human Pose and Shape Estimation CVPR2021
METROEnd-to-End Human Pose and Mesh Reconstruction with Transformers CVPR2021
PyMAF3D Human Pose and Shape Estimation with Pyramidal Mesh Alignment Feedback ICCV2021
PAREPart Attention Regressor for 3D Human Body Estimation ICCV2021
GLAMRGlobal Occlusion-Aware Human Mesh Recovery with Dynamic Motion Filtering CVPR2022
CLIFFCarrying Location Information in Full Frames for Human Mesh Recovery ECCV2022
HMR2.0Humans in 4D: Reconstructing and Tracking Humans with Transformers ICCV2023
SMPLer-XScaling Up Expressive Human Pose and Shape Estimation NeurIPS2023
PyMAF-XWell-aligned Full-body Model Regression from Monocular Images T-PAMI2023
SLAHMRDecoupling Human and Camera Motion from Videos in the Wild CVPR2023
TokenHMRAdvancing Human Mesh Recovery with Tokenized Pose Representation CVPR2024
WHAMReconstructing World-grounded Humans with Accurate 3D Motion CVPR2024
TRAMGlobal Trajectory and Motion of 3D Humans from Videos ECCV2024
NLFNeural Localizer Fields for Continuous 3D Human Pose Estimation NeurIPS2024
GVHMRWorld-Grounded Human Motion Recovery from Monocular Video SIGGRAPH Asia2024
CameraHMRAligning People with Perspective in Human Mesh Recovery 3DV2025
HSMRReconstructing Humans with Biomechanically Accurate Skeleton CVPR2025
BLADESingle-view Body Mesh Learning through Accurate Depth Estimation CVPR2025
PromptHMRPromptable Human Mesh Recovery arXiv2025
SAM-Body4DTraining-Free 4D Human Body Mesh Recovery from Videos arXiv2025
FastHMRAccelerating HMR via Token and Layer Merging arXiv2025
GenHMRGenHMR: Generative Human Mesh Recovery AAAI2025
CoMotionCoMotion: Concurrent Multi-person 3D Motion ICLR2025
GENMOGENMO: A GENarlist Model for Human MOtion ICCV2025
SkelSplatSkelSplat: Robust Multi-view 3D Human Pose Estimation with Differentiable Gaussian Rendering WACV2025
DiffProxyDiffProxy: Multi-View Human Mesh Recovery via Diffusion-Generated Dense Proxies arXiv2026

Egocentric Motion Capture 25+ papers

Model / MethodFull TitleVenueYear
AvatarPoserArticulated full-body pose tracking from sparse motion sensing ECCV2022
EgoEgoEgo-Body Pose Estimation via Ego-Head Pose Estimation CVPR2023
AGRoLAvatars Grow Legs: Smooth Motion from Sparse Tracking with Diffusion CVPR2023
BoDiffusionDiffusing Sparse Observations for Full-Body Motion Synthesis CVPR2023
EgoPoserRobust Real-Time Egocentric Pose Estimation ECCV2024
HMD-PoserOn-Device Real-time Human Motion Tracking from Head-Mounted Devices CVPR2024
EventEgo3D3D Human Motion Capture from Egocentric Event Streams CVPR2024
EgoWholeBodyEgocentric Whole-Body Motion Capture CVPR2024
EgoLMMulti-modal Language Model of Egocentric Motions CVPR2025
EgoAlloEstimating Body and Hand Motion in an Ego-sensed World CVPR2025
Ego4oEgocentric Human Motion from Multi-Modal Input CVPR2025
FRAMEFloor-aligned Representation for Avatar Motion Estimation CVPR2025
UniEgoMotionUnified Model for Egocentric Motion Reconstruction ICCV2025
ECHOEgo-Centric Human-Object Interaction modeling arXiv2025
Fish2Mesh3D Human Mesh Recovery from Fisheye Cameras arXiv2025
EventEgo3D++EventEgo3D++: 3D Human Motion Capture from a Head Mounted Event Camera IJCV2025
EventEgo3D++EventEgo3D++: Egocentric 3D Motion Capture from Monocular Event Cameras with Fisheye Lens arXiv2026
HMD²HMD²: Environment-aware Motion Generation from Single Egocentric Head-Mounted Device 3DV2025
EgoTwinEgoTwin: Dreaming Body and View in First Person arXiv2025
EgoH4EgoH4: The Invisible EgoHand: 3D Hand Forecasting through EgoBody Pose Estimation arXiv2025

Appearance-Rich Human Avatars 19+ papers

Model / MethodFull TitleVenueYear
A-NeRFArticulated Neural Radiance Fields for Learning Human Shape and Appearance NeurIPS2021
SelfReconSelf Reconstruction Your Digital Avatar from Monocular Video CVPR2022
Vid2Avatar3D Avatar Reconstruction from Videos in the Wild CVPR2023
HUGSHuman Gaussian Splats for Real-time Rendering of Animatable Avatars CVPR2024
GaussianAvatarTowards Realistic Human Avatar from a Single Video CVPR2024
Animatable GaussiansLearning Pose-dependent Gaussian Maps for High-fidelity 3D Human Avatar Reconstruction CVPR2024
GPS-GaussianGeneralizable Pixel-wise 3D Gaussian Splatting for Real-time Human Novel View Synthesis CVPR2024
3DGS-AvatarAnimatable Avatars via Deformable 3D Gaussian Splatting CVPR2024
GauSTARGaussian Surface Tracking and Reconstruction for Human Avatars CVPR2025
AHAAHA: Animating Human Avatars in Diverse Scenes with Gaussian Splatting arXiv2025
STG-AvatarSTG-Avatar: Animatable Human Avatars via Spacetime Gaussian IROS2025
LayerGSLayerGS: Decomposition and Inpainting of Layered 3D Human Avatars via 2D Gaussian Splatting arXiv2026
Animated 3DGS AvatarsAnimated 3DGS Avatars in Diverse Scenes with Consistent Lighting and Shadows arXiv2026

Human Interaction (HOI / HSI / HHI) 23+ papers

Model / MethodFull TitleVenueYear
PROXResolving 3D Human Pose Ambiguities with 3D Scene Constraints ICCV2019
PHOSAPerceiving 3D Human-Object Spatial Arrangements from a Single Image in the Wild ECCV2020
BEHAVEDataset and Method for Tracking Human Object Interactions CVPR2022
CHOREContact, Human and Object REconstruction from a Single RGB Image ECCV2022
RICHCapturing and Inferring Dense Full-Body Human-Scene Contact CVPR2022
TRUMANSScaling Up Dynamic Human-Scene Interaction Modeling CVPR2024
BUDDIGenerative Proxemics: A Prior for 3D Social Interaction CVPR2024
AvatarPoseAvatar-guided 3D Pose of Close Human Interaction ECCV2024
Harmony4DVideo Dataset for Close Human Interactions NeurIPS2024
HDMTemplate Free Reconstruction of Human-object Interaction CVPR2024
InterTrackTracking Human Object Interaction without Object Templates 3DV2025
InteractVLM3D Interaction Reasoning from 2D Foundational Models CVPR2025
ODHSROnline Dense 3D Reconstruction of Humans and Scenes CVPR2025
JOSHJoint Optimization for 4D Human-Scene Reconstruction arXiv2025
Ego-Exo4DUnderstanding Skilled Activity from First- and Third-Person Perspectives CVPR2024
HOT3DEgocentric Dataset for 3D Hand and Object Tracking CVPR2025
InterDreamerInterDreamer: Zero-Shot Text to 3D Dynamic Human-Object Interaction NeurIPS2024
MultiPhysMultiPhys: Multi-Person Physics-aware 3D Motion Estimation CVPR2024
CloseIntCloseInt: Closely Interactive Human Reconstruction with Proxemics and Physics-Guided Adaption CVPR2024