Monocular 3D Human Pose Estimation
One stop shop for everything related to 3D human pose estimation from a single RGB image or video.
Please feel free to make pull requests so we can keep this repository as complete and up-to-date as possible.
Table of Contents
Getting Started
Here are some good resources to get started on studying this subfield of computer vision.
Survey papers
Other
3D Human Body Models
Details
List of 3D human body models
Model name | Year | Description | Link |
---|---|---|---|
SCAPE | 2005 | Shape Completion and Animation of People | [Project page] |
BlendSCAPE | 2012 | Coregistration: Simultaneous Alignment and Modeling of Articulated 3D Shape | [Paper PDF] |
SMPL | 2015 | A Skinned Multi-Person Linear Model | [Project page] |
SMPL-H | 2017 | Embodied Hands: Modeling and Capturing Hands and Bodies Together | [PDF page] |
SMPL-X | 2019 | Expressive Body Capture: 3D Hands, Face, and Body from a Single Image | [Project page] |
STAR | 2020 | A Sparse Trained Articulated Human Body Regressor | [Project page] |
GHUM & GHUML | 2020 | Generative 3D Human Shape and Articulated Pose Models | [GitHub page] |
Helpful resources
- SMPL introduction blog: https://khanhha.github.io/posts/SMPL-model-introduction/
Monocular Estimation models
3D Joints
Methods that estimate from a single RGB image or video a human skeleton consisting of 3D joints.
Details
Model name | Year | Description | Temporal | Link |
---|---|---|---|---|
- | - | - | - | - |
3D Mesh
Methods that estimate from a single RGB image or video the whole 3D human body mesh.
Details
Model name | Year | Description | Body model used | Temporal | Link |
---|---|---|---|---|---|
DecoMR | 2020 | 3D Human Mesh Regression with Dense Correspondence | SMPL | ⬜ | [GitHub page] |
VIBE | 2020 | Video Inference for Human Body Pose and Shape Estimation | SMPL | ✅ | [GitHub page] |
- | 2021 | Human Performance Capture from Monocular Video in the Wild | SMPL | ✅ | [PDF] |
SimPoE | 2021 | Simulated Character Control for 3D Human Pose Estimation | ✅ | [PaperswCode Page] |
Extracting Digital Avatars
Extracting from monocular imagary a digital avatar.
Details
Name | Year | Description | Body model used | Link |
---|---|---|---|---|
ICON | 2021 | Implicit Clothed humans Obtained from Normals | [Project page] |
Datasets
All the related datasets for this task. Quickly from memory: AGORA, AMASS, 3DPW, 3.6Human,
Details
Name | Year | Description | Size | Link |
---|---|---|---|---|
3DPW | 2018 | 3D poses in the wild | - | [Project page] |