AI Paper Atlas

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

Unsupervised Deraining: Where Asymmetric Contrastive Learning Meets Self-Similarity

Yi Chang, Yun Guo, Yuntong Ye, Chang‐Feng Yu, Lin Zhu, Xi-Le Zhao, Luxin Yan, Yonghong Tian

TPAMI 2023 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Empirical contrastive visual learning · 31 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 asymmetric contrastive loss (ablation)NLCL · RainCityscapesPSNR: 26.46 -> 27.42Improved
add layer contrastive loss (ablation)ANLCL · RainCityscapesPSNR: 25.83 -> 27.42Improved
add location contrastive loss (ablation)ANLCL · RainCityscapesPSNR: 26.45 -> 27.42Improved
add adversarial loss (ablation)ANLCL · RainCityscapesPSNR: 22.79 -> 27.42Improved
add L1 sparse loss (ablation)ANLCL · RainCityscapesPSNR: 27.07 -> 27.42Improved
non-local sampling for positives and negatives instead of random (ablation)NLCL · RainCityscapesPSNR: 25.83 -> 26.46Improved
neighbour sampling instead of random (ablation)NLCL · RainCityscapesPSNR: 25.83 -> 26.03Improved
use discriminator as contrastive encoder instead of image generator (ablation)NLCL · RainCityscapesPSNR: 24.86 -> 26.46Improved
use 32 positives instead of 8 (256 negatives) (ablation)NLCL · RainCityscapesPSNR: 26.46 -> 25.40Hurt
increase negatives from 64 to 256 (8 positives) (ablation)NLCL · RainCityscapesPSNR: 24.55 -> 26.46Improved
embed ANLCL loss into PReNet (ablation)PReNet · supervised derainingPSNR/SSIM: 35.16 / 0.9762 -> 36.34 / 0.9814Improved
embed ANLCL loss into RCDNetRCDNet · supervised derainingPSNR/SSIM: 36.65 / 0.9805 -> 37.46 / 0.9869Improved

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