Research map / 3D vision / Face and shape
Topologically-Aware Deformation Fields for Single-View 3D Reconstruction
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 |
|---|---|---|---|
| Add learned point features to the deformation field (ablation) | TARS · Pascal3D+ Chair | Chamfer overall: 0.512 -> 0.476 | Helped |
| Add learned point features to the deformation field (ablation) | TARS · Pascal3D+ Car | Chamfer overall: 0.426 -> 0.374 | Helped |
| Add learned point features to the deformation field (ablation) | TARS · Pascal3D+ Airplane | Chamfer overall: 0.568 -> 0.538 | Helped |
| Use a 3D deformation field without point features or SDF correction (ablation) | Single-view implicit reconstruction · ShapeNet Chair | Chamfer overall: 0.333 -> 0.356 | Hurt |
| Add an SDF correction field to the 3D deformation field (ablation) | Single-view implicit reconstruction · ShapeNet Chair | Chamfer overall: 0.356 -> 0.342 | Helped |
| Increase training examples from 500 to 2000 (ablation) | TARS · ShapeNet Chair | Chamfer overall: 0.448 -> 0.385 | Helped |
| Increase training examples from 500 to 2000 (ablation) | SDF-SRN · ShapeNet Chair | Chamfer overall: 0.448 -> 0.386 | Helped |
| Use TARS instead of SoftRas | ShapeNet Chair · ShapeNet Chair | Chamfer overall: 0.523 -> 0.332 | Helped |
| Use TARS instead of SDF-SRN | ShapeNet Car · ShapeNet Car | Chamfer overall: 0.142 -> 0.140 | Helped |
| Use TARS instead of SDF-SRN | ShapeNet Chair · ShapeNet Chair | Chamfer overall: 0.333 -> 0.332 | Helped |
| Use TARS instead of SDF-SRN | ShapeNet Airplane · ShapeNet Airplane | Chamfer overall: 0.173 -> 0.173 | No clear effect |
| Increase point-feature dimension beyond k = 4 (ablation) | TARS deformation field · Pascal3D+ | Chamfer | No clear effect |
Closest papers by meaning
- Canonical 3D Deformer Maps: Unifying parametric and non-parametric methods for dense weakly-supervised category reconstruction (NeurIPS 2020 · similarity 0.89)
- Unsupervised Learning of Dense Shape Correspondence (CVPR 2019 · similarity 0.89)
- Geometry-Aware Network for Non-Rigid Shape Prediction From a Single View (CVPR 2018 · similarity 0.89)
- Deformed Implicit Field: Modeling 3D Shapes With Learned Dense Correspondence (CVPR 2021 · similarity 0.89)
- Pop-Out Motion: 3D-Aware Image Deformation via Learning the Shape Laplacian (CVPR 2022 · similarity 0.88)
- 4Deform: Neural Surface Deformation for Robust Shape Interpolation (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.