Research map / Detection and segmentation / CNN features
Detection Based Part-level Articulated Object Reconstruction from Single RGBD Image
Yuki Kawana, Tatsuya Harada
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 |
|---|---|---|---|
| disable QO, PF and kIoU (KPF module off) (ablation) | Ours-BG · SAPIEN | mAP@90: 38.87 -> 41.09 | Improved |
| add part fusion (PF) on top of query oversampling (ablation) | Ours-BG · SAPIEN | Precision: 52.29 -> 32.65 | Hurt |
| add kinematics-aware IoU (kIoU) on top of QO and PF (ablation) | Ours-BG · SAPIEN | Precision: 32.65 -> 51.64 | Improved |
| add kIoU on top of QO and PF (ablation) | Ours-BG · SAPIEN | mAP@90: 36.48 -> 41.09 | Improved |
| add kIoU on top of QO and PF (ablation) | Ours-BG · SAPIEN | F-Score@90%: 66.24 -> 68.8 | Improved |
| use isotropic scaling instead of anisotropic scale normalization (w/o AS) (ablation) | Ours-BG · SAPIEN | F-Score@90% shape mAP: 68.80 -> 38.83 | Hurt |
| train shape decoder separately without passing shape feature (w/o SF) (ablation) | Ours-BG · SAPIEN | F-Score@90% shape mAP: 68.80 -> 60.49 | Hurt |
| w/o SF on IoU@50% (ablation) | Ours-BG · SAPIEN | IoU@50% shape mAP: 9.80 -> 0.64 | Hurt |
| allocate 25% of decoder layers to refiner R (vary 0% to 75%) (ablation) | Ours-BG · SAPIEN | shape mAP | Improved |
| remove background by using foreground mask input (Ours vs Ours-BG) | Ours · SAPIEN | F-Score@80%: 74.22 -> 74.77 | Improved |
| foreground mask input vs unsegmented | Ours · SAPIEN | IoU@50%: 9.80 -> 11.63 | Improved |
| foreground mask input vs unsegmented on CD@1% | Ours · SAPIEN | CD@1%: 58.61 -> 56.53 | Hurt |
Closest papers by meaning
- CAP-Net: A Unified Network for 6D Pose and Size Estimation of Categorical Articulated Parts from a Single RGB-D Image (CVPR 2025 · similarity 0.90)
- Stable Part Diffusion 4D: Multi-View RGB and Kinematic Parts Video Generation (NeurIPS 2025 · similarity 0.87)
- Hand-held Object Reconstruction from RGB Video with Dynamic Interaction (CVPR 2025 · similarity 0.87)
- Real2Code: Reconstruct Articulated Objects via Code Generation (ICLR 2025 · similarity 0.87)
- Articulate your NeRF: Unsupervised articulated object modeling via conditional view synthesis (NeurIPS 2024 · similarity 0.87)
- ArticulatedGS: Self-supervised Digital Twin Modeling of Articulated Objects using 3D Gaussian Splatting (CVPR 2025 · similarity 0.87)
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.
