AI Paper Atlas

Research map / Learning theory and algorithms / Optimal transport

Sparse Needlets for Lighting Estimation With Spherical Transport Loss

Fangneng Zhan, Changgong Zhang, Wenbo Hu, Shijian Lu, Feiying Ma, Xuansong Xie, Ling Shao

ICCV 2021 · Learning theory and algorithms › Optimal transport · approach: Neural optimal transport · 26 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
use needlets instead of spherical harmonics, both with L2 (ablation)NeedleLight · Laval Indoor HDR (rendered 300 images)STD: 7.14 -> 4.93Improved
add soft thresholding (sparse needlets) with L2 loss (ablation)NeedleLight · Laval Indoor HDR (rendered 300 images)STD: 4.93 -> 4.74Improved
add spherical transport loss to sparse needlets with L2 (ablation)NeedleLight · Laval Indoor HDR (rendered 300 images)STD: 4.74 -> 4.21Improved
add L2 loss to sparse needlets with STL (ablation)NeedleLight · Laval Indoor HDR (rendered 300 images)STD: 4.51 -> 4.21Improved
add STL to spherical harmonic coefficients (ablation)SH basis · Laval Indoor HDR (rendered 300 images)STD: 7.14 -> 7.13No clear effect
add STL to spherical Gaussian coefficients (ablation)SG basis · Laval Indoor HDR (rendered 300 images)STD: 6.01 -> 5.74Improved
add STL to spherical distribution coefficients (ablation)SD basis · Laval Indoor HDR (rendered 300 images)STD: 5.75 -> 5.54Improved
add STL to Haar coefficients (ablation)Haar basis · Laval Indoor HDR (rendered 300 images)STD: 6.07 -> 5.83Improved
add STL to SN(jmax=3) with L2 (ablation)needlets · Laval Indoor HDR (rendered 300 images)STD: 5.32 -> 4.93Improved
use soft instead of hard thresholding (ablation)SN(jmax=3) · Laval Indoor HDR (rendered 300 images)STD: 4.51 -> 4.21Improved
raise needlet order from jmax=1 to jmax=3 (ablation)SN with L2+STL · Laval Indoor HDR (rendered 300 images)STD: 7.18 -> 4.93Improved
raise needlet order from jmax=3 to jmax=4 (ablation)SN with L2+STL · Laval Indoor HDR (rendered 300 images)STD: 4.93 -> 4.96No 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