Research map / Point clouds / Shape and surface completion
Learning Deformable Tetrahedral Meshes for 3D Reconstruction
Read the paperCheck 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 |
|---|---|---|---|
| Add vertex deformation to fixed tetrahedral occupancy (ablation) | DefTet · ShapeNet point-cloud reconstruction | Mean-IoU: 68.98 -> 76.35 | Helped |
| Use the DefTet encoder architecture with OccNet | OccNet · ShapeNet point-cloud reconstruction | Mean-IoU: 62.84 -> 76.93 | Helped |
| Use DefTet instead of OccNet | Point-cloud reconstruction · ShapeNet point-cloud reconstruction | Time (ms): 728.36 -> 61.39 | Helped |
| Apply Marching Tet to the DefTet prediction (ablation) | DefTet · ShapeNet single-image reconstruction with 2D supervision | Mean Chamfer distance: 3.6 -> 3.1 | Helped |
| Add vertex deformation rather than keep a fixed tetrahedral grid (ablation) | DefTet · ShapeNet single-image reconstruction with 2D supervision | Mean Chamfer distance: 4.5 -> 3.6 | Helped |
| Add vertex deformation while using Marching Tet (ablation) | FixedTet + MTet · ShapeNet single-image reconstruction with 2D supervision | Mean Chamfer distance: 3.5 -> 3.1 | Helped |
| Use amortized learning for meshing instead of optimization alone (ablation) | DefTet tetrahedral meshing · Chair tetrahedral meshing | Run Time (sec.): 1956.80 -> 49.13 | Helped |
| Use amortized learning instead of optimization alone (ablation) | DefTet tetrahedral meshing · Chair tetrahedral meshing | AMIPS: 3.37 -> 3.61 | Hurt |
| Use amortized learning instead of optimization alone (ablation) | DefTet tetrahedral meshing · Chair tetrahedral meshing | Chamfer (e-3): 4.21 -> 4.27 | Hurt |
Closest papers by meaning
- Deep Marching Tetrahedra: a Hybrid Representation for High-Resolution 3D Shape Synthesis (NeurIPS 2021 · similarity 0.90)
- Learning Dynamic Tetrahedra for High-Quality Talking Head Synthesis (CVPR 2024 · similarity 0.88)
- TetSphere Splatting: Representing High-Quality Geometry with Lagrangian Volumetric Meshes (ICLR 2025 · similarity 0.87)
- Tetrahedron Splatting for 3D Generation (NeurIPS 2024 · similarity 0.87)
- Diffeomorphic Mesh Deformation via Efficient Optimal Transport for Cortical Surface Reconstruction (ICLR 2024 · similarity 0.86)
- Sparc3D: Sparse Representation and Construction for High-Resolution 3D Shapes Modeling (NeurIPS 2025 · 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.