AI Paper Atlas

Research map / 3D vision / 3D generation

SiPhy: Single-Image Physical Property Reasoning

Hoang Le, Joonwoo Kwon, Elkhan Ismayilzada, Yufei Zhang, Zijun Cui

ECCV 2026 · 3D vision › 3D generation · approach: Learning physics from video · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
add heaviness-aware thickness refinement (HAT) (ablation)SiPhy · ABO-500-heavyADE: 18.86 -> 15.59Improved
add heaviness-aware thickness refinement (HAT) (ablation)SiPhy · ABO-500-heavyMnRE: 0.57 -> 0.65Improved
add heaviness-aware thickness refinement (HAT) (ablation)SiPhy · ABO-500-lightADE: 2.07 -> 3.57Hurt
add heaviness-aware thickness refinement (HAT) (ablation)SiPhy · ABO-500 (all)MnRE: 0.53 -> 0.58Improved
depth-adaptive 2D sampling instead of uniform spacing (ablation)SiPhy · ABO-500 massMnRE: 0.39 -> 0.58Improved
part-based contrastive alignment instead of averaging SAM mask features (ablation)SiPhy · ABO-500 massMnRE: 0.54 -> 0.58Improved
single-view SiPhy instead of multi-view NeRF2PhysicsSiPhy · ABO-500 massMnRE: 0.55 -> 0.58Improved
single-view SiPhy instead of multi-view PUGSSiPhy · ABO-500 massMnRE: 0.30 -> 0.58Improved
single-view SiPhy instead of multi-view GaussianPropertySiPhy · ABO-500 material segmentationmIoU: 0.29 -> 0.25Hurt
single-view SiPhy instead of multi-view NeRF2PhysicsSiPhy · PhysXNet-100 densityMAE: 2044 -> 1315Improved
single-view SiPhy instead of multi-view PUGSSiPhy · PhysXNet-100 densityMAE: 1297 vs 1315No clear effect
single-view SiPhy instead of multi-view PUGSSiPhy · PhysXNet-100 Young's modulusMAE: 68 -> 52Improved

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy