Research map / 3D vision / Human pose estimation
"Category-level Object Detection, Pose Estimation and Reconstruction from Stereo Images"
Chuanrui Zhang, Yonggen Ling, Minglei Lu, Minghan Qin, Haoqian Wang
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 |
|---|---|---|---|
| replace 2D position embedding with stereo-aware 3D position embedding (ablation) | CODERS · TOD (bottle) | 5deg 2cm: 70.8 -> 73.4 | Improved |
| add position embedding with 3D information versus none (ablation) | CODERS · TOD (bottle) | 5deg 2cm: 53.5 -> 73.4 | Improved |
| use stereo input instead of monocular (ablation) | CODERS · TOD (bottle) | 3D75: 16.1 -> 31.5 | Improved |
| increase decoder layers from 1 to 6 (ablation) | CODERS · TOD (bottle) | 5deg 2cm: 29.6 -> 73.4 | Improved |
| increase decoder layers from 6 to 8 (ablation) | CODERS · TOD (bottle) | 5deg 2cm: 73.4 -> 73.5 | No clear effect |
| add contrastive loss to shape encoder (ablation) | CODERS · Reconstruction (bottle) | Chamfer distance x100: 1.411 -> 0.542 | Improved |
| add shape head to pose training (ablation) | CODERS · TOD (bottle) | 5deg 2cm: 61.1 -> 73.4 | Improved |
| add pose head to shape training (ablation) | CODERS · TOD (bottle) | Chamfer distance: 0.223 -> 0.201 | Improved |
| one-stage end-to-end stereo model versus prior stereo method StereoPose | TOD bottle · TOD (bottle) | 3D50: 22.2 -> 99.5 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- KeyPose — KeyPose: Multi-View 3D Labeling and Keypoint Estimation for Transparent Objects (CVPR 2020) · better on 1 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
- Unsupervised Learning of Category-Level 3D Pose from Object-Centric Videos (CVPR 2024 · similarity 0.86)
- Vision Foundation Model Enables Generalizable Object Pose Estimation (NeurIPS 2024 · similarity 0.86)
- CatFormer: Category-Level 6D Object Pose Estimation with Transformer (AAAI 2024 · similarity 0.86)
- Universal Features Guided Zero-Shot Category-Level Object Pose Estimation (AAAI 2025 · similarity 0.85)
- GenCOPE: Syn2Real Generalized Category-Level Object Pose Estimation for Robotic Picking (NeurIPS 2026 · similarity 0.85)
- DenseMatcher: Learning 3D Semantic Correspondence for Category-Level Manipulation from a Single Demo (ICLR 2025 · similarity 0.85)
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.
