Research map / Point clouds / Shape and surface completion
A Skeleton-Bridged Deep Learning Approach for Generating Meshes of Complex Topologies From Single RGB Images
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 |
|---|---|---|---|
| Use skeleton-bridged, stage-wise reconstruction instead of the strongest mean-CD baseline, AtlasNet | Single-view ShapeNet reconstruction · ShapeNet test (mean) | CD (× 10^3): 1.832 -> 1.422 | Helped |
| Use skeleton-bridged, stage-wise reconstruction instead of the strongest mean-EMD baseline, Pixel2Mesh | Single-view ShapeNet reconstruction · ShapeNet test (mean) | EMD (× 0.01): 7.490 -> 5.934 | Helped |
| Add parallel curve and surface skeleton streams rather than point-only fitting (ablation) | Skeleton inference · ShapeNet skeleton inference | CD (× 10^3): 1.185 -> 1.103 | Helped |
| Add the square stream to line-only fitting (ablation) | Skeleton inference · ShapeNet skeleton inference | CD (× 10^3): 1.649 -> 1.103 | Helped |
| Add the line stream to square-only fitting (ablation) | Skeleton inference · ShapeNet skeleton inference | CD (× 10^3): 1.185 -> 1.103 | Helped |
| Use separate MLP streams rather than one MLP for line and square primitives (ablation) | Skeleton inference · ShapeNet skeleton inference | CD (× 10^3): 1.252 -> 1.103 | Helped |
| Add Laplacian smoothness loss (ablation) | Skeleton inference · ShapeNet skeleton inference | CD (× 10^3): 1.621 -> 1.103 | Helped |
| Remove skeleton inference and generate a high-resolution volume directly from the image (ablation) | Full reconstruction pipeline · Thin-structure objects | topology reconstruction | Hurt |
| Remove voxel-based volume correction and directly convert the inferred skeleton to a mesh (ablation) | Full reconstruction pipeline · Single-view reconstruction | final mesh quality | Hurt |
Closest papers by meaning
- Point2Skeleton: Learning Skeletal Representations from Point Clouds (CVPR 2021 · similarity 0.89)
- Skeleton-bridged Point Completion: From Global Inference to Local Adjustment (NeurIPS 2020 · similarity 0.88)
- Learning Unsupervised Hierarchical Part Decomposition of 3D Objects From a Single RGB Image (CVPR 2020 · similarity 0.88)
- Im2Struct: Recovering 3D Shape Structure From a Single RGB Image (CVPR 2018 · similarity 0.88)
- Geometry-Aware Network for Non-Rigid Shape Prediction From a Single View (CVPR 2018 · similarity 0.87)
- SKDream: Controllable Multi-view and 3D Generation with Arbitrary Skeletons (CVPR 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.