AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

Semi-Supervised Learning of Semantic Correspondence With Pseudo-Labels

Jiwon Kim, Kwangrok Ryoo, Junyoung Seo, Gyuseong Lee, Daehwan Kim, Hansang Cho, Seungryong Kim

CVPR 2022 · Label-efficient and robust learning › Semi-supervised learning · approach: Confidence-based pseudo-label selection · 28 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
add pseudo-label-based semi-supervised training to CATsCATs · PF-PASCALPCK@0.1: 92.6 -> 93.5Improved
add pseudo-label-based semi-supervised training to CATsCATs · PF-WillowPCK@0.1: 79.2 -> 82.1Improved
add pseudo-label-based semi-supervised training to CATsCATs · SPair-71kPCK@0.1: 49.9 -> 50.7Improved
add foreground object masking to pseudo-label selection (ablation)SemiMatch · PF-PASCALPCK@0.05: 73.9 -> 80.1Improved
add forward-backward consistency masking (ablation)SemiMatch · PF-WillowPCK@0.05: 52.9 -> 54.0Improved
add uncertainty-based threshold weighting (ablation)SemiMatch · PF-PASCALPCK@0.05: 72.3 -> 80.1Improved
replace ordinary CutOut with weak keypoint-guided CutOut (ablation)SemiMatch Base · PF-WillowPCK@0.05: 51.8 -> 54.0Improved
use strong rather than weak keypoint-guided CutOut (ablation)SemiMatch Base · PF-PASCALPCK@0.1: 93.6 -> 93.4No clear effect
remove the 20-epoch supervised warm-up (ablation)SemiMatch · PF-PASCALPCK@0.1: 93.6 -> 93.5No clear effect
replace contrastive loss with AEPE loss (ablation)SemiMatch · PF-PASCALPCK@0.1: 93.6 -> 92.7Hurt

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