Research map / 3D vision / 3D generation
SiPhy: Single-Image Physical Property Reasoning
Hoang Le, Joonwoo Kwon, Elkhan Ismayilzada, Yufei Zhang, Zijun Cui
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 heaviness-aware thickness refinement (HAT) (ablation) | SiPhy · ABO-500-heavy | ADE: 18.86 -> 15.59 | Improved |
| add heaviness-aware thickness refinement (HAT) (ablation) | SiPhy · ABO-500-heavy | MnRE: 0.57 -> 0.65 | Improved |
| add heaviness-aware thickness refinement (HAT) (ablation) | SiPhy · ABO-500-light | ADE: 2.07 -> 3.57 | Hurt |
| add heaviness-aware thickness refinement (HAT) (ablation) | SiPhy · ABO-500 (all) | MnRE: 0.53 -> 0.58 | Improved |
| depth-adaptive 2D sampling instead of uniform spacing (ablation) | SiPhy · ABO-500 mass | MnRE: 0.39 -> 0.58 | Improved |
| part-based contrastive alignment instead of averaging SAM mask features (ablation) | SiPhy · ABO-500 mass | MnRE: 0.54 -> 0.58 | Improved |
| single-view SiPhy instead of multi-view NeRF2Physics | SiPhy · ABO-500 mass | MnRE: 0.55 -> 0.58 | Improved |
| single-view SiPhy instead of multi-view PUGS | SiPhy · ABO-500 mass | MnRE: 0.30 -> 0.58 | Improved |
| single-view SiPhy instead of multi-view GaussianProperty | SiPhy · ABO-500 material segmentation | mIoU: 0.29 -> 0.25 | Hurt |
| single-view SiPhy instead of multi-view NeRF2Physics | SiPhy · PhysXNet-100 density | MAE: 2044 -> 1315 | Improved |
| single-view SiPhy instead of multi-view PUGS | SiPhy · PhysXNet-100 density | MAE: 1297 vs 1315 | No clear effect |
| single-view SiPhy instead of multi-view PUGS | SiPhy · PhysXNet-100 Young's modulus | MAE: 68 -> 52 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- GaussianProperty — GaussianProperty: Integrating Physical Properties to 3D Gaussians with LMMs (ICCV 2025) · better on 2 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Physical Property Understanding from Language-Embedded Feature Fields (CVPR 2024 · similarity 0.87)
- ContPhy: Continuum Physical Concept Learning and Reasoning from Videos (ICML 2024 · similarity 0.86)
- Physically Compatible 3D Object Modeling from a Single Image (NeurIPS 2024 · similarity 0.85)
- UniPhy: Learning a Unified Constitutive Model for Inverse Physics Simulation (CVPR 2025 · similarity 0.84)
- A General Protocol to Probe Large Vision Models for 3D Physical Understanding (NeurIPS 2024 · similarity 0.84)
- Beyond Static Vision: Scene Dynamic Field Unlocks Intuitive Physics Understanding in Multi-modal Large Language Models (ICLR 2026 · similarity 0.84)
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.
