AI Paper Atlas

Research map / Architectures and efficiency / CNN design

End-To-End Projector Photometric Compensation

Bingyao Huang, Haibin Ling

CVPR 2019 · Architectures and efficiency › CNN design · approach: Dense matching and estimation CNNs · 39 citations (OpenAlex)

Check 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
include the camera-captured surface image as an input (ablation)CompenNet · Projector compensation benchmark, 24 setups, 500 training imagesPSNR: 20.2618 -> 21.7998Improved
include the camera-captured surface image as an input (ablation)CompenNet · Projector compensation benchmark, 24 setups, 500 training imagesRMSE: 0.1698 -> 0.1425Improved
include the camera-captured surface image as an input (ablation)CompenNet · Projector compensation benchmark, 24 setups, 500 training imagesSSIM: 0.7209 -> 0.7523Improved
pre-train on 500 images projected onto a white surface (ablation)CompenNet · Projector compensation benchmark, 24 setups, 32 training pairs and 500 iterationsPSNR: 19.6767 -> 20.3423Improved
pre-train on 500 images projected onto a white surface (ablation)CompenNet · Projector compensation benchmark, 24 setups, 32 training pairs and 500 iterationsRMSE: 0.1863 -> 0.1688Improved
pre-train on 500 images projected onto a white surface (ablation)CompenNet · Projector compensation benchmark, 24 setups, 32 training pairs and 500 iterationsSSIM: 0.6788 -> 0.7165Improved
use L1 plus SSIM loss instead of pixel-wise L1 loss (ablation)CompenNet · Projector compensation benchmark, 24 setups, 500 training imagesPSNR: 21.1782 -> 21.7998Improved
use L1 plus SSIM loss instead of pixel-wise L2 loss (ablation)CompenNet · Projector compensation benchmark, 24 setups, 500 training imagesPSNR: 20.7927 -> 21.7998Improved
use L1 plus SSIM loss instead of SSIM loss alone (ablation)CompenNet · Projector compensation benchmark, 24 setups, 500 training imagesPSNR: 21.0134 -> 21.7998Improved
use L1 plus SSIM loss instead of SSIM loss alone (ablation)CompenNet · Projector compensation benchmark, 24 setups, 500 training imagesSSIM: 0.7591 -> 0.7523Hurt
increase training images from 125 to 500 (ablation)CompenNet · Projector compensation benchmark, 24 setupsPSNR: 21.0542 -> 21.7998Improved
increase training images from 125 to 500 (ablation)CompenNet · Projector compensation benchmark, 24 setupsSSIM: 0.7314 -> 0.7523Improved

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