Research map / 3D vision / Face and shape
Joint Learning of 3D Shape Retrieval and Deformation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Jointly train retrieval and deformation instead of using pretrained deformation with deformation-aware retrieval | Retrieval-and-deformation framework · Image-to-mesh, Chair / Table / Cabinet | Chamfer distance (× 10^-2): 1.216 -> 1.005; 1.621 -> 0.970; 1.333 -> 1.220 | Helped |
| Jointly train retrieval and deformation instead of using pretrained deformation with deformation-aware retrieval | Retrieval-and-deformation framework · Points-to-mesh, Chair / Table / Cabinet | Chamfer distance (× 10^-2): 0.965 -> 0.763; 1.561 -> 0.696; 0.829 -> 0.715 | Helped |
| Add biased source sampling instead of uniform sampling during joint training (ablation) | Joint retrieval-and-deformation training · Image-to-mesh, Chair / Table / Cabinet | Chamfer distance (× 10^-2): 1.118 -> 1.005; 1.486 -> 0.970; 1.318 -> 1.220 | Helped |
| Add biased source sampling instead of uniform sampling during joint training (ablation) | Joint retrieval-and-deformation training · Points-to-mesh, Chair / Table / Cabinet | Chamfer distance (× 10^-2): 0.998 -> 0.763; 1.502 -> 0.696; 0.767 -> 0.715 | Helped |
| Add inner deformation optimization | Ours · Image-to-mesh, Chair / Table / Cabinet | Chamfer distance (× 10^-2): 1.005 -> 0.976; 0.970 -> 0.935; 1.220 -> 1.141 | Helped |
| Add direct optimization after the predicted deformation | Ours · Points-to-mesh, Chair / Table / Cabinet | Chamfer distance (× 10^-2): 0.763 -> 0.504; 0.696 -> 0.414; 0.715 -> 0.494 | Helped |
| Jointly train the deformation module on retrieval-relevant shape pairs instead of random pairs | Deformation module · Image-to-mesh with oracle retrieval, Chair / Table / Cabinet | Chamfer distance (× 10^-2): 0.748 -> 0.681; 0.702 -> 0.584; 0.706 -> 0.675 | Helped |
| Jointly train the deformation module on retrieval-relevant shape pairs instead of random pairs | Deformation module · Points-to-mesh with oracle retrieval, Chair / Table / Cabinet | Chamfer distance (× 10^-2): 0.712 -> 0.643; 0.703 -> 0.564; 0.549 -> 0.494 | Helped |
| Use joint training with Neural Cages rather than pretrained Neural Cages | Neural Cages deformation module · Points-to-mesh, Chair / Table / Cabinet | Chamfer distance (× 10^-2): 0.480 -> 0.476; 0.575 -> 0.411; 0.589 -> 0.538 | Mixed |
| Use joint training with auto-segmented sources instead of the listed baselines | Structure-aware retrieval-and-deformation · Auto-segmented sources, Chair / Table | Chamfer distance (× 10^-2): 1.118 -> 0.990; 1.409 -> 1.166 | Helped |
| Remove connectivity constraints (ablation) | Joint retrieval-and-deformation framework · Points-to-mesh, Chair / Table / Cabinet | Chamfer distance (× 10^-2) | No clear effect |
| Increase the source database size while using joint training | Chair retrieval-and-deformation framework · Points-to-mesh, source databases of 50, 100, 200, 400, and 800 models | Chamfer distance (× 10^-2): 0.823; 0.803; 0.748; 0.784; 0.734 | Mixed |
Closest papers by meaning
- KP-RED: Exploiting Semantic Keypoints for Joint 3D Shape Retrieval and Deformation (CVPR 2024 · similarity 0.89)
- Reduced Representation of Deformation Fields for Effective Non-rigid Shape Matching (NeurIPS 2022 · similarity 0.89)
- Neural Shape Deformation Priors (NeurIPS 2022 · similarity 0.88)
- ShapeMatcher: Self-Supervised Joint Shape Canonicalization Segmentation Retrieval and Deformation (CVPR 2024 · similarity 0.87)
- Learning Partonomic 3D Reconstruction from Image Collections (CVPR 2025 · similarity 0.87)
- Neural Pose Representation Learning for Generating and Transferring Non-Rigid Object Poses (NeurIPS 2024 · similarity 0.86)
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.