Research map / Point clouds / Shape and surface completion
SALD: Sign Agnostic Learning with Derivatives
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 a sign-agnostic derivative term to SAL loss | SAL implicit shape reconstruction · Shape approximation, including L-shaped curves | surface detail and sharp-feature preservation | Helped |
| Use SALD with an auto-decoder rather than a VAE | SALD · ShapeNet test-car reconstruction | visual reconstruction detail | Helped |
| Use SALD instead of SAL | ShapeNet shape-space learning · ShapeNet Sofas | mean Chamfer distance: 0.704 -> 0.207 | Helped |
| Use SALD instead of SAL | ShapeNet shape-space learning · ShapeNet Chairs | mean Chamfer distance: 0.494 -> 0.281 | Helped |
| Use SALD instead of SAL | ShapeNet shape-space learning · ShapeNet Tables | mean Chamfer distance: 0.543 -> 0.408 | Helped |
| Use SALD instead of SAL | ShapeNet shape-space learning · ShapeNet Planes | mean Chamfer distance: 0.429 -> 0.098 | Helped |
| Use SALD instead of SAL | ShapeNet shape-space learning · ShapeNet Lamps | mean Chamfer distance: 4.913 -> 0.506 | Helped |
| Use SALD instead of IGR | D-Faust shape reconstruction · D-Faust | mean dC(reconstruction,registration): 3.806 -> 0.489 | Helped |
| Use SALD instead of IGR | D-Faust shape reconstruction · D-Faust | mean dC(scan,reconstruction): 0.241 -> 0.397 | Hurt |
| Use SALD instead of SAL | D-Faust shape reconstruction · D-Faust | mean dC(registration,reconstruction): 0.418 -> 0.428 | Hurt |
| Add the sign-agnostic derivative term | SAL implicit representation · D-Faust and ShapeNet raw geometry | reconstruction quality and detail | Mixed |
Closest papers by meaning
- SAL: Sign Agnostic Learning of Shapes From Raw Data (CVPR 2020 · similarity 0.94)
- 3PSDF: Three-Pole Signed Distance Function for Learning Surfaces With Arbitrary Topologies (CVPR 2022 · similarity 0.87)
- DeepSDF: Learning Continuous Signed Distance Functions for Shape Representation (CVPR 2019 · similarity 0.87)
- Few-Shot Unsupervised Implicit Neural Shape Representation Learning with Spatial Adversaries (ICML 2024 · similarity 0.86)
- Small Steps and Level Sets: Fitting Neural Surface Models with Point Guidance (CVPR 2024 · similarity 0.85)
- LISR: Learning Linear 3D Implicit Surface Representation Using Compactly Supported Radial Basis Functions (AAAI 2024 · 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.