AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

CVF-SID: Cyclic Multi-Variate Function for Self-Supervised Image Denoising by Disentangling Noise From Image

Reyhaneh Neshatavar, Mohsen Yavartanoo, Sanghyun Son, Kyoung Mu Lee

CVPR 2022 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Pretext-task visual pre-training · 108 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
train directly on the evaluation dataset rather than SIDD-MediumCVF-SID · SIDD validationPSNR: 34.51 -> 34.67Improved
double-train by applying two successive models instead of oneCVF-SID · SIDD validationPSNR: 34.67 -> 34.81Improved
add identity loss (ablation)CVF-SID · SIDD validationPSNR: 33.10 -> 34.24Improved
add zero loss after identity loss (ablation)CVF-SID · SIDD validationPSNR: 34.24 -> 34.29Improved
add regularization loss after zero loss (ablation)CVF-SID · SIDD validationPSNR: 34.29 -> 34.43Improved
add augmentation set A (ablation)CVF-SID · SIDD validationPSNR: 34.43 -> 34.48Improved
add augmentation set B to set A (ablation)CVF-SID · SIDD validationPSNR: 34.48 -> 34.67Improved
vary γ from 0.25 through 1.50 (ablation)CVF-SID · SIDD validationPSNR: 34.45; 34.46; 34.67; 34.66Mixed
cascade three models rather than two (ablation)CVF-SID · SIDD validationPSNR: 34.81 -> 34.84Improved
cascade four models rather than three (ablation)CVF-SID · SIDD validationPSNR: 34.84 -> 34.84No clear effect
increase augmentation loss weight from 0.1 to 1 (ablation)CVF-SID · SIDD validationPSNR: 34.67 -> 34.55Hurt

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