- Single-Person 2D Pose Estimation
- 3D Mesh Recovery from video
- 3D People Tracking
- Multi-Person 3D Mesh Recovery
- Single-Person 3D Mesh Recovery
- Multi-Person 2D Pose Estimation
- Multi-Person 3D Pose Estimation
- Backbone
• PyMAF-X: Towards Well-aligned Full-body Model Regression from Monocular Images - [code] [paper] - Arxiv, PyMAF-X
Hongwen Zhang, Yating Tian, Yuxiang Zhang, Mengcheng Li, Liang An, Zhenan Sun, Yebin Liu
• Accurate 3D Hand Pose Estimation for Whole-Body 3D Human Mesh Estimation - [code] [paper] - CVPRW, Hand4Whole
Gyeongsik Moon, Hongsuk Choi, Kyoung Mu Lee
Yao Feng, Vasileios Choutas, Timo Bolkart, Dimitrios Tzionas, Michael J. Black
• FrankMocap: Fast Monocular 3D Hand and Body Motion Capture by Regression and Integration - [code] [paper] - ICCV workshop, FrankMocap
Ye Yuan, Umar Iqbal, Pavlo Molchanov, Kris Kitani, Jan Kautz
• Monocular Real-time Full Body Capture with Inter-part Correlations - [paper] - CVPR 21, Zhou et al
Yuxiao Zhou, Marc Habermann, Ikhsanul Habibie, Ayush Tewari, Christian Theobalt, Feng Xu
• Whole-Body Human Pose Estimation in the Wild - [code] [paper] - ECCV, COCO-WholeBody, (only Keypoint)
Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping
• Monocular Expressive Body Regression through Body-Driven Attention - [code] [paper] - ECCV, Expose
Vasileios Choutas, Georgios Pavlakos, Timo Bolkart, Dimitrios Tzionas, Michael J. Black
• DOPE: Distillation Of Part Experts for whole-body 3D pose estimation in the wild - [code] [paper] - ECCV, DOPE, (only Keypoint)
Philippe Weinzaepfel, Romain Brégier, Hadrien Combaluzier, Vincent Leroy, Grégory Rogez
• Expressive Body Capture: 3D Hands, Face, and Body from a Single Image -[code] [paper] - CVPR 19, SMPL-X
Pavlakos, Georgios and Choutas, Vasileios and Ghorbani, Nima and Bolkart, Timo and Osman, Ahmed A. A. and Tzionas, Dimitrios and Black, Michael J.
• SimCC: a Simple Coordinate Classification Perspective for Human Pose Estimation - [code] [paper] - ECCV 22, SimCC
• GLAMR: Global Occlusion-Aware Human Mesh Recovery with Dynamic Cameras - [code] [paper] - CVPR 22, GLAMR
Ye Yuan, Umar Iqbal, Pavlo Molchanov, Kris Kitani, Jan Kautz
• Capturing Humans in Motion: Temporal-Attentive 3D Human Pose and Shape Estimation from Monocular Video - [paper] - CVPR 22, MPS-Net
Wen-Li Wei, Jen-Chun Lin, Tyng-Luh Liu, Hong-Yuan Mark Liao
• Beyond Static Features for Temporally Consistent 3D Human Pose and Shape from a Video - [code] [paper] - CVPR 21, TCMR
Hongsuk Choi, Gyeongsik Moon, Ju Yong Chang, Kyoung Mu Lee
Muhammed Kocabas, Nikos Athanasiou, Michael J. Black
Jathushan Rajasegaran, Georgios Pavlakos, Angjoo Kanazawa, Jitendra Malik
• TesseTrack: End-to-End Learnable Multi-Person Articulated 3D Pose Tracking - [paper] - CVPR 21, TesseTrack
N Dinesh Reddy, Laurent Guigues, Leonid Pishchulin, Jayan Eledath, Srinivasa G. Narasimhan
Jathushan Rajasegaran, Georgios Pavlakos, Angjoo Kanazawa, Jitendra Malik
• Putting People in their Place: Monocular Regression of 3D People in Depth - [code] [paper] [preview] - CVPR 22, BEV
Yu Sun, Wu Liu, Qian Bao, Yili Fu, Tao Mei, Michael J. Black
• Learning to Estimate Robust 3D Human Mesh from In-the-Wild Crowded Scenes - [code] [paper] - CVPR 22, 3DCrowdNet
Hongsuk Choi, Gyeongsik Moon, JoonKyu Park, Kyoung Mu Lee
Yu Sun, Qian Bao, Wu Liu, Yili Fu, Michael J. Black, Tao Mei
• HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation - [code] [paper] - CVPR 21, HybrIK
Jiefeng Li, Chao Xu, Zhicun Chen, Siyuan Bian, Lixin Yang, Cewu Lu
Kevin Lin, Lijuan Wang, Zicheng Liu
Kevin Lin, Lijuan Wang, Zicheng Liu
• Distribution-Aware Single-Stage Models for Multi-Person 3D Pose Estimation - [paper] - CVPR 22, DAS
Zitian Wang, Xuecheng Nie, Xiaochao Qu, Yunpeng Chen, Si Liu
• Learning Local-Global Contextual Adaptation for Multi-Person Pose Estimation - [code] [paper] - CVPR 22, LOGO-CAP
Nan Xue, Tianfu Wu, Gui-Song Xia, Liangpei Zhang
Dahu Shi1, Xing Wei2, Liangqi Li, Ye Ren, Wenming Tan
• Lite Pose: Efficient Architecture Design for 2D Human Pose Estimation - [code] [paper] - CVPR 22, Lite Pose
Yihan Wang, Muyang Li, Han Cai, Wei-Ming Chen, Song Han
• Contextual Instance Decoupling for Robust Multi-Person Pose Estimation - [code] [paper] - CVPR 22, CID
Dongkai Wang, Shiliang Zhang
• Location-Free Human Pose Estimation - [paper] - CVPR 22
Xixia Xu, Yingguo Gao, Ke Yan, Xue Lin, Qi Zou
• Rethinking Keypoint Representations: Modeling Keypoints and Poses as Objects for Multi-Person Human Pose Estimation - [code] [paper] - Arxiv 21.11, KAPAO
William McNally, Kanav Vats, Alexander Wong, John McPhee
• Bottom-Up Human Pose Estimation Via Disentangled Keypoint Regression - [code] [paper] - CVPR 21, DEKR
Zigang Geng, Ke Sun, Bin Xiao, Zhaoxiang Zhang, Jingdong Wang
• OpenPifPaf: Composite Fields for Semantic Keypoint Detection and Spatio-Temporal Association - [code] [paper] - Arxiv 21.03, OpenPifPaf
Sven Kreiss, Lorenzo Bertoni, Alexandre Alahi
Jiefeng Li, Siyuan Bian, Ailing Zeng, Can Wang, Bo Pang, Wentao Liu, Cewu Lu
• Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation - [code] [paper] - ICCV 21, MIPNet
Rawal Khirodkar, Visesh Chari, Amit Agrawal, Ambrish Tyagi
• Robust Pose Estimation in Crowded Scenes with Direct Pose-Level Inference - [code] [paper] - NeurIPS 21, PINet
Dongkai Wang, Shiliang Zhang, Gang Hua
• HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation - [code] [paper] - CVPR 20, HigherHRNet
Bowen Cheng, Bin Xiao, Jingdong Wang, Honghui Shi, Thomas S. Huang, Lei Zhang
• Distribution-Aware Coordinate Representation for Human Pose Estimation - [code] [paper] - CVPR 20, DARK
Feng Zhang, Xiatian Zhu, Hanbin Dai, Mao Ye, Ce Zhu
Sven Kreiss, Lorenzo Bertoni, Alexandre Alahi
• Single-Stage is Enough: Multi-Person Absolute 3D Pose Estimation - [paper] - CVPR 22, DRM
Lei Jin, Chenyang Xu, Xiaojuan Wang, Yabo Xiao, Yandong Guo, Xuecheng Nie, Jian Zhao
• ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation - [code] [paper] - Arxiv 22, ViTPose
Yufei Xu, Jing Zhang, Qiming Zhang, Dacheng Tao
• HRFormer: High-Resolution Transformer for Dense Prediction - [code] [paper] - NeurIPS 21, HRFormer
Yuhui Yuan, Rao Fu, Lang Huang, Weihong Lin, Chao Zhang, Xilin Chen, Jingdong Wang
| Dataset | # Frames | # Scenes | # Subjects | # Subjects Per Frame | In-the-wild | Mesh Type | Mesh Annotation Source |
|---|---|---|---|---|---|---|---|
| SURREAL | 6.5M | 2,607 | 145 | 1 | - | SMPL | |
| GTA-Human | 1.4M | - | >600 | - | - | SMPL | |
| AGORA | 17K | - | >350 | 5~15 | - | SMPL-X | |
| THUman2.0 | - | - | ~200 | 1 | - | SMPL-X | |
| MultiHuman | - | - | ~50 | 1~3 | - | SMPL-X |
| Dataset | # Frames | # Scenes | # Subjects | # Subjects Per Frame | In-the-wild | Mesh Type | Mesh Annotation Source |
|---|---|---|---|---|---|---|---|
| HumanEva | 80K | 1 | 4 | 1 | - | - | |
| Human3.6M | 3.6M | 1 | 11 | 1 | - | SMPL-X | NeuralAnnot |
| 3DPW | >51K | 60 | 7 | 1~2 | yes | SMPL-X | NeuralAnnot |
| Dataset | # Frames | # Scenes | # Subjects | # Subjects Per Frame | In-the-wild | Mesh Type | Mesh Annotation Source |
|---|---|---|---|---|---|---|---|
| CMU Panoptic | 1.5M | 1 | 40 | 3~8 | - | - | |
| MPI-INF-3DHP | >1.3M | 1 | 8 | 1 | yes | SMPL-X | NeuralAnnot |
| MuCo-3DHP | 200K | 1 | 8 | 1~4 | - | - | |
| MuPoTs-3D | >8K | 20 | 8 | 3 | yes | - | |
| MannequinChallenge | 24,428 | 567 | 742 | 5 | yes | SMPL | |
| 3DOH50K | 51,600 | 1 | - | 1 | - | SMPL | |
| Mirrored-Human | 1.8M | >200 | >200 | >=1 | yes | SMPL | |
| MTC | 834K | 1 | 40 | 1 | - | - | |
| EHF | 100 | 1 | 1 | 1 | - | SMPL-X | |
| HUMBI | 17.3M | 1 | 772 | 1 | - | SMPL | |
| ZJU-MoCap | - | 1 | 9 | 1 | - | SMPL-X |
| Dataset | # Frames | # Scenes | # Subjects | # Subjects Per Frame | In-the-wild | Mesh Type | Mesh Annotation Source |
|---|---|---|---|---|---|---|---|
| LSP | 2K | - | - | 1 | yes | SMPL | |
| LSP-Extended | 10K | - | - | 1 | yes | SMPL | |
| PennAction | 77K | 2,326 | 2,326 | 1 | yes | SMPL | |
| MSCOCO | 38K | - | - | >=1 | yes | SMPL | |
| MPII | 24,920 | 3,913 | >40K | >=1 | yes | SMPL | |
| UP-3D | 8,515 | - | - | 1 | yes | SMPL | |
| PoseTrack | 66,374 | 550 | 550 | >1 | yes | SMPL-X | NeuralAnnot |
| SSP-3D | 311 | 62 | 62 | 1 | yes | SMPL | |
| OCHuman | 4,731 | - | 8110 | >1 | yes | SMPL | |
| MTP | 3,731 | - | 148 | 1 | yes | SMPL-X |
- borrowed from KAPAO
