AI Paper Atlas

Research map / Image generation / Image restoration

CompenNet++: End-to-End Full Projector Compensation

Bingyao Huang, Haibin Ling

ICCV 2019 · Image generation › Image restoration · approach: Image harmonization and color transformation · 34 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 the grid refinement network to CompenNet++ (ablation)CompenNet++ · Full projector compensation benchmark (#Train=500)PSNR: 20.9172 -> 21.1127Improved
add the grid refinement network to CompenNet++ (ablation)CompenNet++ · Full projector compensation benchmark (#Train=48)PSNR: 19.4139 -> 19.8552Improved
add the grid refinement network to CompenNet++ (ablation)CompenNet++ · Full projector compensation benchmark (#Train=500)SSIM: 0.7113 -> 0.7269Improved
use end-to-end CompenNet++ instead of two-step CompenNet with structured light warpingprojector full compensation · Full projector compensation benchmark (#Train=500)PSNR: 20.9552 -> 21.1127Improved
use end-to-end CompenNet++ instead of two-step CompenNet with structured light warpingprojector full compensation · Full projector compensation benchmark (#Train=48)PSNR: 20.2023 -> 19.8552Hurt
replace CompenNet++ with general image-to-image model Pix2pix with structured light warpingprojector full compensation · Full projector compensation benchmark (#Train=500)PSNR: 21.1127 -> 19.4160Hurt
train TPS on colorful textured images instead of plain color imagesTPS with structured light · Full projector compensation benchmarkPSNR: 16.7271 -> 17.9746Improved
use the fast variant (1,000 iterations, batch 24, 48 images) of CompenNet++CompenNet++ · Full projector compensation benchmarkPSNR: 19.8552 -> 19.9696Mixed
use the faster variant (500 iterations, batch 16, 48 images) of CompenNet++CompenNet++ · Full projector compensation benchmark (#Train=48)PSNR: 19.8552 -> 19.2536Hurt
increase training set size from 48 to 500 imagesCompenNet++ · Full projector compensation benchmarkPSNR: 19.8552 -> 21.1127Improved

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