Research map / 3D vision / Human pose estimation
Human shape and pose regression
Estimates body shape and pose with mesh regression, probabilistic and whole-body models.
80 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Human pose estimation, keyword cluster “body · motion · human mesh”
State History Benchmarks Tried Checklist Preprints Who Export
Where it stands
Still moving The best reported result is still improving on 1 of 6 tracked benchmarks.
17 papers in 2023–24 and 12 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 6.3% to 4.0%.
Open questions the data points to
- 3DPW (MPJPE): best went from 72 (up to 2024) to 64.5 in 2025 — Reconstructing Close Human Interaction with Appearance and Proxemics Reasoning (CVPR 2025 · 3 citations)
No longer reported by papers here: Human3.6M (MPJPE, last in 2024); 3DPW (PA-MPJPE, last in 2023); Human3.6M (PA-MPJPE, last in 2023); 3DPW (PVE, last in 2023); MPI-INF-3DHP (PCK, last in 2024).
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Recurrent Attention Models for Depth-Based Person Identification (CVPR 2016 · 170 citations)
- 2017 · Learning From Synthetic Humans (CVPR 2017 · 974 citations)
- 2018 · Learning to Estimate 3D Human Pose and Shape From a Single Color Image (CVPR 2018 · 658 citations)
- 2019 · Expressive Body Capture: 3D Hands, Face, and Body From a Single Image (CVPR 2019 · 1,796 citations)
- 2020 · VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020 · 1,076 citations)
- 2021 · HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation (CVPR 2021 · 399 citations)
- 2022 · Pose-Guided Feature Disentangling for Occluded Person Re-identification Based on Transformer (AAAI 2022 · 231 citations)
- 2023 · BEDLAM: A Synthetic Dataset of Bodies Exhibiting Detailed Lifelike Animated Motion (CVPR 2023 · 141 citations)
- 2024 · Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimation (NeurIPS 2024 · 17 citations)
- 2025 · IDOL: Instant Photorealistic 3D Human Creation from a Single Image (CVPR 2025 · 31 citations)
- 2026 · Bringing Your Portrait to 3D Presence (CVPR 2026 · 0 citations)
Earliest papers here
- Recurrent Attention Models for Depth-Based Person Identification (CVPR 2016 · 170 citations)
- Improving Person Re-Identification via Pose-Aware Multi-Shot Matching (CVPR 2016 · 107 citations)
- Learning From Synthetic Humans (CVPR 2017 · 974 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
3DPW · MPJPE (lower is better)
11 papers, 2020–2025 · still improving
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 75.8 | 3D Multi-bodies: Fitting Sets of Plausible 3D Human Models to Ambiguous Image Data (NeurIPS 2020 · 5 citations) |
| 2021 | 74.1 | HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation (CVPR 2021 · 399 citations) |
| 2023 | 72 | BEDLAM: A Synthetic Dataset of Bodies Exhibiting Detailed Lifelike Animated Motion (CVPR 2023 · 141 citations) |
| 2025 | 64.5 | Reconstructing Close Human Interaction with Appearance and Proxemics Reasoning (CVPR 2025 · 3 citations) |
Human3.6M · MPJPE (lower is better)
7 papers, 2020–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 65.6 | VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020 · 1,076 citations) |
| 2021 | 54.4 | HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation (CVPR 2021 · 399 citations) |
| 2024 | 39.7 | Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimation (NeurIPS 2024 · 17 citations) |
3DPW · PA-MPJPE (lower is better)
6 papers, 2020–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 51.9 | VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020 · 1,076 citations) |
| 2021 | 45 | HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation (CVPR 2021 · 399 citations) |
| 2023 | 42 | Capturing the Motion of Every Joint: 3D Human Pose and Shape Estimation with Independent Tokens (ICLR 2023 · 4 citations) |
Human3.6M · PA-MPJPE (lower is better)
6 papers, 2020–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 41.4 | VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020 · 1,076 citations) |
| 2021 | 33.6 | HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation (CVPR 2021 · 399 citations) |
3DPW · PVE (lower is better)
5 papers, 2020–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 99.1 | VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020 · 1,076 citations) |
| 2021 | 86.5 | HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation (CVPR 2021 · 399 citations) |
| 2023 | 85 | BEDLAM: A Synthetic Dataset of Bodies Exhibiting Detailed Lifelike Animated Motion (CVPR 2023 · 141 citations) |
MPI-INF-3DHP · PCK (higher is better)
4 papers, 2020–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 89.3 | VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020 · 1,076 citations) |
| 2024 | 97.5 | Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimation (NeurIPS 2024 · 17 citations) |
What has been tried
75 papers here report 798 measured changes (a component added, removed or swapped, compared with a baseline): helped 77%, no clear effect 4%, hurt 17%, mixed 2%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add HuGe100K to training | Human reconstruction test set · MSE | 0.017 -> 0.008 | Helped | IDOL: Instant Photorealistic 3D Human Creation from a Single Image (CVPR 2025 · 31 citations) |
| use GPS positional encoding instead of plain XYZ coordinates | SSP-3D · mIoU | 0.777 -> 0.789 | Helped | Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimation (NeurIPS 2024 · 17 citations) |
| Regress Euler angles instead of continuous rotation representation | MOYO · MPJPE | 124.0 -> 137.1 | Hurt | Reconstructing Humans with a Biomechanically Accurate Skeleton (CVPR 2025 · 14 citations) |
| add pose decomposition (PD) | LaFAN1, 30 transition frames · L2Q | 0.59 -> 0.57 | Helped | A Unified Masked Autoencoder with Patchified Skeletons for Motion Synthesis (AAAI 2024 · 10 citations) |
| Add CtrlAni3D pretraining | Animal3D · PA-MPJPE | 127.8 -> 121.9 | Helped | AniMer: Animal Pose and Shape Estimation Using Family Aware Transformer (CVPR 2025 · 7 citations) |
| add the SLE projection of 3D Fourier features into 2D | CustomHuman and THuman3.0 · f-score | 31.624 -> 42.425; 35.086 -> 46.091 | Helped | MultiGO: Towards Multi-level Geometry Learning for Monocular 3D Textured Human Reconstruction (CVPR 2025 · 6 citations) |
| detach human-loss gradients from scene and camera parameters | EgoHumans · W-MPJPE (m) | 1.04 -> 3.94 | Hurt | Reconstructing People, Places, and Cameras (CVPR 2025 · 6 citations) |
| initialize CroCo-Body pre-training from CroCo weights rather than random weights | 3DPW test · PA-MPJPE | 60.5 -> 57.5 | Helped | Cross-view and Cross-pose Completion for 3D Human Understanding (CVPR 2024 · 6 citations) |
| add appearance branch (3DGS constraint) to optimization | Hi4D · MPJPE | 60.68 -> 59.06 | Helped | Reconstructing Close Human Interaction with Appearance and Proxemics Reasoning (CVPR 2025 · 3 citations) |
| add the template branch to image branches | THuman2.0 ablation, 1 input · FID | 66.53 -> 57.78 | Helped | NoPo-Avatar: Generalizable and Animatable Avatars from Sparse Inputs without Human Poses (NeurIPS 2025 · 0 citations) |
| Add the Bypass Network and LDL to SMPLer-X | Fashion-body · MAE across chest, waist, hip, wrist, and shoulder width (cm) | Chest: 6.34 -> 3.07; waist: 9.84 -> 3.91; hip: 11.05 -> 6.95; wrist: 1.13 -> 0.41; shoulder width: 4.57 -> 2.03 | Helped | A Focused Human Body Model for Accurate Anthropometric Measurements Extraction (CVPR 2025) |
What reviewers will expect
Datasets and baselines named in the 8 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- COCO · 2 of 8 papers
- HuGe100K · 2 of 8 papers
- THuman 2.1 · 2 of 8 papers
- THuman2.0 · 2 of 8 papers
- 3DPW · 2 of 8 papers
Baselines
- HMR2.0 · 2 of 8 papers
- GTA · 2 of 8 papers
- SIFU · 2 of 8 papers
- CLIFF · 2 of 8 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 29 papers; from OpenAlex and the papers' affiliations).
- Max Planck Institute for Intelligent Systems · 3 papers
- University of California, Berkeley · 2 papers
- Southeast University · 2 papers
- Nanyang Technological University · 2 papers
- SenseTime · 2 papers
Use it
Download BibTeX of 35 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Physics and sensor-based motion capture (125 papers)
- Human mesh recovery from images (58 papers)
- Human motion prediction (38 papers)
Elsewhere in the topic
- Learned camera pose estimation (96 papers)
- Category-level object pose estimation (92 papers)
- Monocular 2D-to-3D pose lifting (90 papers)
- Depth-based hand pose regression (84 papers)
- Transformer-based multi-view 3D pose (83 papers)
- Temporal video pose estimation (76 papers)
- Instance-level 6D object pose (74 papers)
- Category-level pose estimation (54 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).