Research map / 3D vision / Human pose estimation
Category-level pose estimation
Estimate object pose and part motion from point clouds.
54 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2018–2026) · an approach in Human pose estimation, keyword cluster “object pose · 6d · hand”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
25 papers in 2023–24 and 12 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 9.2% to 4.0%.
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
- 2018 · Hand PointNet: 3D Hand Pose Estimation Using Point Sets (CVPR 2018 · 328 citations)
- 2019 · Embodied Question Answering in Photorealistic Environments With Point Cloud Perception (CVPR 2019 · 134 citations)
- 2020 · GraspNet-1Billion: A Large-Scale Benchmark for General Object Grasping (CVPR 2020 · 667 citations)
- 2021 · MultiBodySync: Multi-Body Segmentation and Motion Estimation via 3D Scan Synchronization (CVPR 2021 · 36 citations)
- 2022 · SAR-Net: Shape Alignment and Recovery Network for Category-Level 6D Object Pose and Size Estimation (CVPR 2022 · 95 citations)
- 2023 · UniDexGrasp: Universal Robotic Dexterous Grasping via Learning Diverse Proposal Generation and Goal-Conditioned Policy (CVPR 2023 · 113 citations)
- 2024 · Dexterous Grasp Transformer (CVPR 2024 · 29 citations)
- 2025 · CAP-Net: A Unified Network for 6D Pose and Size Estimation of Categorical Articulated Parts from a Single RGB-D Image (CVPR 2025 · 2 citations)
- 2026 · Fourier Features Let Agents Learn High Precision Policies with Imitation Learning (ICML 2026 · 0 citations)
Earliest papers here
- Hand PointNet: 3D Hand Pose Estimation Using Point Sets (CVPR 2018 · 328 citations)
- Unsupervised Articulated Skeleton Extraction From Point Set Sequences Captured by a Single Depth Camera (AAAI 2018 · 17 citations)
- Embodied Question Answering in Photorealistic Environments With Point Cloud Perception (CVPR 2019 · 134 citations)
What has been tried
46 papers here report 496 measured changes (a component added, removed or swapped, compared with a baseline): helped 74%, no clear effect 5%, hurt 19%, mixed 3%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add AB-TTA | DexGraspNet · Q1 | 0.0278 -> 0.0515 | Helped | Dexterous Grasp Transformer (CVPR 2024 · 29 citations) |
| add jitter during t-patch construction | DFAUST · top-1 frame accuracy | 74.73 -> 80.49 | Helped | 3DInAction: Understanding Human Actions in 3D Point Clouds (CVPR 2024 · 24 citations) |
| Add joint-wise condition, local condition, joint indicator, kinematic correspondence, and multiple hypotheses | DexYCB · mean joint error | 9.17 mm -> 7.66 mm | Helped | HandDiff: 3D Hand Pose Estimation with Diffusion on Image-Point Cloud (CVPR 2024 · 23 citations) |
| add CatFormer's coarse deformation module | REAL275 · 5°2cm | 32.5 -> 47.7 | Helped | CatFormer: Category-Level 6D Object Pose Estimation with Transformer (AAAI 2024 · 14 citations) |
| Add the Global Perception module | View-S2HGD · penetration volume (cm3) | 11.54 -> 6.58 | Helped | Single-View Scene Point Cloud Human Grasp Generation (CVPR 2024 · 12 citations) |
| use post-sampling instead of pre-sampling | ManiSkill2 tasks, ACT · Mean S.R. | 0.21 -> 0.41 SpUNet; 0.31 -> 0.45 PointNet | Helped | Point Cloud Matters: Rethinking the Impact of Different Observation Spaces on Robot Learning (NeurIPS 2024 · 10 citations) |
| Use LiDAR, RGB, and Event inputs instead of RGB alone | RELI11D · MPJPE | 62.71 -> 49.19 | Helped | RELI11D: A Comprehensive Multimodal Human Motion Dataset and Method (CVPR 2024 · 8 citations) |
| use 8 keypoints per part instead of 4 | PartNet-Mobility Laptop · rotation error | Mixed | KPA-Tracker: Towards Robust and Real-Time Category-Level Articulated Object 6D Pose Tracking (AAAI 2024 · 6 citations) | |
| rotate the world frame to stabilize P1P2L | Generic synthetic · mean rotation error | 0.00050 -> 9.0e-09 | Helped | Efficient Solution of Point-Line Absolute Pose (CVPR 2024 · 3 citations) |
| Add AdaFormer | KITTI · Success | 45.9 -> 54.0 | Helped | Towards Category Unification of 3D Single Object Tracking on Point Clouds (ICLR 2024 · 3 citations) |
| Train the semantic, instance-offset, and NPCS heads jointly rather than independently | RGBD-Art · rotation error Re | 15.30 -> 10.39 | Helped | CAP-Net: A Unified Network for 6D Pose and Size Estimation of Categorical Articulated Parts from a Single RGB-D Image (CVPR 2025 · 2 citations) |
| add closer-aligned-retrace exploration instead of coverage-based exploration | RoomR · Success (%) | 13.1 -> 14.2 | Helped | A Category Agnostic Model for Visual Rearrangment (CVPR 2024 · 2 citations) |
What reviewers will expect
Datasets and baselines named in the 17 papers of this approach from 2024–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- CAMERA25 · 2 of 17 papers
- REAL275 · 2 of 17 papers
- PartNet-Mobility · 2 of 17 papers
- RobotArm · 2 of 17 papers
Baselines
- NOCS · 2 of 17 papers
- SPD · 2 of 17 papers
- SGPA · 2 of 17 papers
- CR-Net · 2 of 17 papers
- FS-Net · 2 of 17 papers
- DualPoseNet · 2 of 17 papers
- SAR-Net · 2 of 17 papers
- GPV-Pose · 2 of 17 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 37 papers; from OpenAlex and the papers' affiliations).
- Stanford University · 4 papers
- Shanghai Jiao Tong University · 4 papers
- Tsinghua University · 3 papers
- Peking University · 3 papers
- Beijing Institute of Technology · 2 papers
- University of Science and Technology of China · 2 papers
- Hefei University of Technology · 2 papers
- University of California, San Diego · 2 papers
Use it
Download BibTeX of 38 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
- Depth-based hand pose regression (84 papers)
- Learned camera pose estimation (96 papers)
- Category-level object pose estimation (92 papers)
- Instance-level 6D object pose (74 papers)
Elsewhere in the topic
- Physics and sensor-based motion capture (125 papers)
- Monocular 2D-to-3D pose lifting (90 papers)
- Transformer-based multi-view 3D pose (83 papers)
- Human shape and pose regression (80 papers)
- Temporal video pose estimation (76 papers)
- Human mesh recovery from images (58 papers)
- Multi-person pose estimation (51 papers)
- Heatmap and coordinate pose encoding (50 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).