Research map / 3D vision / Face and shape
DeepSDF: Learning Continuous Signed Distance Functions for Shape Representation
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 continuous SDF prediction with latent-code conditioning | DeepSDF · ShapeNet cars, known shapes | CD mean (×10^3): 0.167 -> 0.084 | Helped |
| Use continuous SDF prediction with latent-code conditioning | DeepSDF · ShapeNet chair, unknown shapes | CD mean: 0.368 -> 0.204 | Helped |
| Use continuous SDF prediction with latent-code conditioning | DeepSDF · ShapeNet plane, unknown shapes | CD mean: 0.216 -> 0.143 | Helped |
| Use continuous SDF prediction with latent-code conditioning | DeepSDF · ShapeNet table, unknown shapes | CD mean: 0.328 -> 0.553 | Hurt |
| Use continuous SDF prediction with latent-code conditioning | DeepSDF · ShapeNet lamp, unknown shapes | CD mean: 1.182 -> 0.832 | Helped |
| Use continuous SDF prediction with latent-code conditioning | DeepSDF · ShapeNet sofa, unknown shapes | CD mean: 0.411 -> 0.132 | Helped |
| Infer the shape code from partial range scans using the learned decoder prior | DeepSDF · ShapeNet chair, partial range scans | CD mean: 2.83 -> 2.11 | Helped |
| Infer the shape code from partial range scans using the learned decoder prior | DeepSDF · ShapeNet plane, partial range scans | CD mean: 2.19 -> 1.16 | Helped |
| Infer the shape code from partial range scans using the learned decoder prior | DeepSDF · ShapeNet sofa, partial range scans | CD mean: 2.18 -> 1.59 | Helped |
| Represent shape classes with a continuous decoder rather than a voxel or octree model | DeepSDF · ShapeNet shape representation | Model size (GB): 0.54 -> 0.0074 | Helped |
| Add input-vector skip connections to intermediate layers (ablation) | DeepSDF · ShapeNet chairs, training SDF regression | training loss | Helped |
| Use 8 layers rather than 4 (ablation) | DeepSDF · ShapeNet chairs, SDF regression | test error: 9.7 -> 5.7 | Helped |
Closest papers by meaning
- Local Deep Implicit Functions for 3D Shape (CVPR 2020 · similarity 0.91)
- Few-Shot Unsupervised Implicit Neural Shape Representation Learning with Spatial Adversaries (ICML 2024 · similarity 0.90)
- Sensing Surface Patches in Volume Rendering for Inferring Signed Distance Functions (AAAI 2025 · similarity 0.90)
- A Lightweight UDF Learning Framework for 3D Reconstruction Based on Local Shape Functions (CVPR 2025 · similarity 0.89)
- Small Steps and Level Sets: Fitting Neural Surface Models with Point Guidance (CVPR 2024 · similarity 0.89)
- Mosaic-SDF for 3D Generative Models (CVPR 2024 · similarity 0.88)
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.