Research map / 3D vision / Human pose estimation
Category-level object pose estimation
Estimates object pose for unseen instances within categories using correspondence learning and universal features.
92 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · 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.
22 papers in 2023–24 and 35 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 8.1% to 11.7%.
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 · Exploiting View-Specific Appearance Similarities Across Classes for Zero-Shot Pose Prediction: A Metric Learning Approach (AAAI 2016 · 19 citations)
- 2017 · Deep MANTA: A Coarse-To-Fine Many-Task Network for Joint 2D and 3D Vehicle Analysis From Monocular Image (CVPR 2017 · 484 citations)
- 2018 · RotationNet: Joint Object Categorization and Pose Estimation Using Multiviews From Unsupervised Viewpoints (CVPR 2018 · 506 citations)
- 2019 · Strike (With) a Pose: Neural Networks Are Easily Fooled by Strange Poses of Familiar Objects (CVPR 2019 · 259 citations)
- 2020 · Category-Level Articulated Object Pose Estimation (CVPR 2020 · 194 citations)
- 2021 · Single-View Robot Pose and Joint Angle Estimation via Render & Compare (CVPR 2021 · 42 citations)
- 2022 · OnePose: One-Shot Object Pose Estimation Without CAD Models (CVPR 2022 · 169 citations)
- 2023 · HS-Pose: Hybrid Scope Feature Extraction for Category-Level Object Pose Estimation (CVPR 2023 · 78 citations)
- 2024 · GigaPose: Fast and Robust Novel Object Pose Estimation via One Correspondence (CVPR 2024 · 87 citations)
- 2025 · Gaze-LLE: Gaze Target Estimation via Large-Scale Learned Encoders (CVPR 2025 · 26 citations)
- 2026 · Exploring Category-level Articulated Object Pose Tracking on SE(3) Manifolds (AAAI 2026 · 2 citations)
Earliest papers here
- Exploiting View-Specific Appearance Similarities Across Classes for Zero-Shot Pose Prediction: A Metric Learning Approach (AAAI 2016 · 19 citations)
- Deep MANTA: A Coarse-To-Fine Many-Task Network for Joint 2D and 3D Vehicle Analysis From Monocular Image (CVPR 2017 · 484 citations)
- POSEidon: Face-From-Depth for Driver Pose Estimation (CVPR 2017 · 169 citations)
What has been tried
68 papers here report 758 measured changes (a component added, removed or swapped, compared with a baseline): helped 72%, no clear effect 5%, hurt 21%, 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 |
|---|---|---|---|---|
| fine-tune Fae instead of using DINOv2 features without fine-tuning | LM-O, T-LESS, YCB-V mean · AR | 19.0 -> 27.9 | Helped | GigaPose: Fast and Robust Novel Object Pose Estimation via One Correspondence (CVPR 2024 · 87 citations) |
| add semantic features | NOCS-REAL275 · 5°2cm | 51.1 -> 56.2 | Helped | SecondPose: SE(3)-Consistent Dual-Stream Feature Fusion for Category-Level Pose Estimation (CVPR 2024 · 37 citations) |
| Add head prompting | GazeFollow test · AUC | 0.926 -> 0.956 | Helped | Gaze-LLE: Gaze Target Estimation via Large-Scale Learned Encoders (CVPR 2025 · 26 citations) |
| Use object masks during NVS training and evaluation | WildRGB-D single-scene NVS · PSNR | 23.03/1.50 -> 34.65/4.44; 23.84/1.60 -> 35.60/4.51; 23.67/2.07 -> 35.65/5.20 | Helped | RGBD Objects in the Wild: Scaling Real-World 3D Object Learning from RGB-D Videos (CVPR 2024 · 24 citations) |
| Train on 10% rather than the full training set | RealEstate10K · PSNR | 25.01 -> 19.32 | Hurt | You See it, You Got it: Learning 3D Creation on Pose-Free Videos at Scale (CVPR 2025 · 21 citations) |
| use nonlinear aggregation instead of vanilla aggregation | LM, unseen objects · Acc15 | 98.3 -> 98.2 | No clear effect | Object Pose Estimation via the Aggregation of Diffusion Features (CVPR 2024 · 17 citations) |
| replace cuboid proxy shape with ellipsoid | LINEMOD · ADD(s) | 60.9 -> 58.0 | Hurt | MFOS: Model-Free & One-Shot Object Pose Estimation (AAAI 2024 · 13 citations) |
| add instance geometric prior only | HouseCat6D · 5°2cm | 21.3 -> 22.2 | Helped | GCE-Pose: Global Context Enhancement for Category-level Object Pose Estimation (CVPR 2025 · 12 citations) |
| Add random rotational camera perturbation data augmentation during training | Olsson test scenes, inference + BA · average reprojection error (px) | 1.92 -> 0.85 | Helped | Learning Structure-from-Motion with Graph Attention Networks (CVPR 2024 · 10 citations) |
| Use WCV instead of pose regression | CO3D · angular error | 31.78 -> 19.95 | Helped | DVMNet: Computing Relative Pose for Unseen Objects Beyond Hypotheses (CVPR 2024 · 8 citations) |
| Add RDFGrad-based projection | Pose denoising · ∆q +m2m | 0.171 -> 0.170 | No clear effect | NRDF: Neural Riemannian Distance Fields for Learning Articulated Pose Priors (CVPR 2024 · 8 citations) |
| use the smaller model instead of Ours (L) | GSO · R. error | 2.46 -> 13.08 | Hurt | PF-LRM: Pose-Free Large Reconstruction Model for Joint Pose and Shape Prediction (ICLR 2024 · 7 citations) |
What reviewers will expect
Datasets and baselines named in the 20 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- REAL275 · 6 of 20 papers
- CAMERA25 · 6 of 20 papers
- HouseCat6D · 4 of 20 papers
- Wild6D · 3 of 20 papers
- Objaverse · 2 of 20 papers
- NOCS REAL275 · 2 of 20 papers
- CO3D · 2 of 20 papers
Baselines
- DPDN · 5 of 20 papers
- VI-Net · 5 of 20 papers
- SPD · 4 of 20 papers
- SGPA · 4 of 20 papers
- SecondPose · 4 of 20 papers
- DualPoseNet · 3 of 20 papers
- NOCS · 3 of 20 papers
- AG-Pose · 3 of 20 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 57 papers; from OpenAlex and the papers' affiliations).
- Tsinghua University · 4 papers
- EPFL · 3 papers
- Peking University · 3 papers
- Chinese Academy of Sciences · 3 papers
- University of Science and Technology of China · 3 papers
- Google · 3 papers
- Technical University of Munich · 3 papers
- University of Chinese Academy of Sciences · 2 papers
- The University of Hong Kong · 2 papers
- Peng Cheng Laboratory · 2 papers
Use it
Download BibTeX of 39 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)
- Instance-level 6D object pose (74 papers)
- Category-level pose estimation (54 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).