AI Paper Atlas

Research map / Detection and segmentation / CNN features

Decoupled Multi-Task Learning With Cyclical Self-Regulation for Face Parsing

Qingping Zheng, Jiankang Deng, Zheng Zhu, Ying Li, Stefanos Zafeiriou

CVPR 2022 · Detection and segmentation › CNN features · approach: Local features and context in CNNs · 51 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 DDGCN context embedding without pooling (ablation)Baseline face parsing model · HelenOverall F1: 92.11 -> 92.56Improved
Add DDGCN context embedding without pooling (ablation)Baseline face parsing model · CelebAMask-HQMean F1: 84.34 -> 85.33Improved
Add DDGCN context embedding without pooling (ablation)Baseline face parsing model · LaPaMean F1: 89.84 -> 92.10Improved
Add decoupled multi-task edge learning (ablation)DDGCN face parsing model · HelenOverall F1: 92.56 -> 93.75Improved
Add decoupled multi-task edge learning (ablation)DDGCN face parsing model · CelebAMask-HQMean F1: 85.33 -> 85.98Improved
Add decoupled multi-task edge learning (ablation)DDGCN face parsing model · LaPaMean F1: 92.10 -> 92.32Improved
Add cyclical self-regulation (ablation)DDGCN + DML face parsing model · HelenOverall F1: 93.75 -> 93.78Improved
Add cyclical self-regulation (ablation)DDGCN + DML face parsing model · CelebAMask-HQMean F1: 85.98 -> 86.07Improved
Add cyclical self-regulation (ablation)DDGCN + DML face parsing model · LaPaMean F1: 92.32 -> 92.38Improved
Replace pooled context with pooling-removed PSP (ablation)Baseline face parsing model · HelenOverall F1: 92.20 -> 92.37Improved
Replace pooled context with pooling-removed DGCNet (ablation)Baseline face parsing model · HelenOverall F1: 92.41 -> 92.45Improved
Add binary edge branch (ablation)DDGCN face parsing model · CelebAMask-HQMean F1: 84.34 -> 85.58Improved

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