Research map / 3D vision / Human pose estimation
3D-Object Perception Transformer (3PT)
Agastya Kalra, Tim Salzmann, Guy Stoppi, Dmitrii Marin, Rishav Agarwal, Vage Taamazyan, Martin Bokeloh, Stefan Hinterstoisser, Anton Boykov, Alberto Dall'Olio, Pravin Dangol, Kartik Venkataraman and 1 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase training 3D-model vocabulary from 35k to 110k CADs (ablation) | 3PT-D · BOP-H3 and BOP-Industrial mean | AP: 49.7 -> 54.8 | Improved |
| train 300k instead of 100k iterations (S=3) (ablation) | 3PT-D · BOP mean | AP: 54.8 -> 56.4 | Improved |
| replace MUSE ensemble outputs with 3PT classification, scoring and regression (ablation) | MUSE · HANDAL | AP: 35.7 -> 43.7 | Improved |
| train end-to-end 3D-conditioned detector instead of patching MUSE (ablation) | MUSE · HANDAL | AP: 43.7 -> 53.6 | Improved |
| apply 3PT multi-view RGB refinement to FRTPose predictions (ablation) | FRTPose · BOP-Industrial | AP-mm: 68.3 -> 78.3 | Improved |
| add 3PT-D detection and initialization to full pipeline (ablation) | 3PT-R on FRTPose · BOP-Industrial | AP-mm: 78.3 -> 80.6 | Improved |
| apply 3PT-R refinement to Co-Op predictions (ablation) | Co-Op · BOP-H3 single-view | AP: 46.4 -> 47.8 | Improved |
| use 3PT-D initial poses with K=1 (ablation) | 3PT-R · BOP-H3 single-view | AP: 47.8 -> 55.1 | Improved |
| use K=5 instead of K=1 orientation hypotheses (ablation) | 3PT · BOP-H3 single-view | AP: 55.1 -> 58.7 | Improved |
| use K=8 instead of K=1 rotation hypotheses (ablation) | 3PT · BOP-Industrial | AP-mm: 71.1 -> 80.9 | Improved |
| use 3 refinement iterations instead of 1 (ablation) | 3PT-R · BOP-Industrial | AP-mm: 77.2 -> 80.9 | Improved |
| provide known object classes a priori (ablation) | 3PT · BOP-Industrial | AP-mm: 80.6 -> 80.9 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- GigaPose — GigaPose: Fast and Robust Novel Object Pose Estimation via One Correspondence (CVPR 2024) · better on 2 of 2 shared results
- FreeZeV2 — FreeZe: Training-free zero-shot 6D pose estimation with geometric and vision foundation models (ECCV 2024) · better on 0 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ImageNet3D: Towards General-Purpose Object-Level 3D Understanding (NeurIPS 2024 · similarity 0.85)
- Multiple View Geometry Transformers for 3D Human Pose Estimation (CVPR 2024 · similarity 0.84)
- CatFormer: Category-Level 6D Object Pose Estimation with Transformer (AAAI 2024 · similarity 0.83)
- ImOV3D: Learning Open Vocabulary Point Clouds 3D Object Detection from Only 2D Images (NeurIPS 2024 · similarity 0.83)
- Towards In-the-wild 3D Plane Reconstruction from a Single Image (CVPR 2025 · similarity 0.83)
- OpenAD: Open-World Autonomous Driving Benchmark for 3D Object Detection (NeurIPS 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
