AI Paper Atlas

Research map / 3D vision / Depth and stereo

Shape and Spatially-Varying Reflectance Estimation from Virtual Exemplars

Zhuo Hui, Aswin C. Sankaranarayanan

TPAMI 2016 · 3D vision › Depth and stereo · approach: Photometric stereo

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
use coarse-to-fine search instead of brute forcevirtual exemplar normal estimation · synthetic normals, 0.5 deg samplingtime: 74.1s -> 0.41sImproved
use coarse-to-fine search instead of brute forcevirtual exemplar normal estimation · synthetic normals, 0.5 deg samplingangular error: 0.42 -> 0.82Hurt
add gradient descent refinement (ablation)coarse-to-fine · synthetic normals, 0.5 deg samplingangular error: 0.82 -> 0.38Improved
add gradient descent refinement (ablation)brute force · synthetic normals, 0.5 deg samplingangular error: 0.42 -> 0.21Improved
add gradient descent refinement (ablation)coarse-to-fine · synthetic normals, 10 deg samplingangular error: 7.07 -> 4.10Improved
add gradient descent to coarse-to-fine vs Shi et al.normal estimation · synthetic MERL, 200 imagesangular error: 0.88 -> 0.58Improved
gradient descent vs competing methodsnormal estimation · MERL materials (253 images)angular error: better in 88 out of 100 materialsImproved
gradient descent vs benchmarked methodsnormal estimation · photometric stereo benchmarkmean angular error: best on 8 of 10 objects, e.g. harvest 25.44 -> 15.81Improved
add low-rank prior to per-pixel BRDF estimation (ablation)BRDF estimation · synthetic spatially-varying sphererelative BRDF errorImproved
use dictionary of mismatched material type (ablation)normal estimation · MERL 10 test materials, 253 imagesangular errorHurt
use leave-one-out dictionary of 99 atoms vs smaller dictionaries (ablation)normal estimation · MERL 10 test materialsangular error: 99 atoms vs 12 to 23 atomsImproved
reduce dictionary to 10-90 randomly chosen atoms (ablation)gradient descent normals · MERL, 253 imagesangular error: less than 3 deg for most materialsNo clear effect

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