AI Paper Atlas

Research map / Label-efficient and robust learning / Knowledge distillation

Online Knowledge Distillation for Efficient Pose Estimation

Zheng Li, Jingwen Ye, Mingli Song, Ying Huang, Zhigeng Pan

ICCV 2021 · Label-efficient and robust learning › Knowledge distillation · approach: Few-shot and metric distillation · 124 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
train with OKDHP online distillation2-Stack HG · MPII valPCKh@0.5: 88.6 -> 89.2Improved
train with OKDHP online distillation4-Stack HG · MPII valPCKh@0.5: 89.2 -> 90.0Improved
train with OKDHP online distillation8-Stack HG · MPII valPCKh@0.5: 90.2 -> 90.6Improved
train with OKDHP online distillation4-Stack HG · COCO val2017AP: 73.6 -> 74.8Improved
use pixel-wise KL divergence instead of MSE as distillation loss (ablation)OKDHP 4-Stack HG · MPII valPCKh@0.5: 89.5 -> 90.0Improved
use FAU instead of simple mean aggregation of branch heatmaps (ablation)OKDHP 4-Stack HG · MPII valPCKh@0.5: 89.7 -> 90.0Improved
use FAU instead of ONE-style gate module (ablation)OKDHP 4-Stack HG · MPII valPCKh@0.5: 89.8 -> 90.0Improved
increase number of branches from 2 to 5 (ablation)OKDHP 2-Stack HG · MPII valPCKh@0.5: 89.2 -> 89.4Improved
increase individual HG modules per branch from 1 to 4 (ablation)OKDHP 4-Stack HG · MPII valPCKh@0.5: 89.8 -> 90.2Improved
vary loss weight beta from 0.5 to 5 (ablation)OKDHP 2-Stack HG · MPII valPCKh@0.5: 89.24 -> 89.18No clear effect
use unbalanced branches (6- and 8-stack auxiliaries) instead of balanced (ablation)OKDHP 4-Stack HG · MPII valPCKh@0.5: 90.0 -> 90.2Improved
use OKDHP instead of offline FPD distillation4-Stack HG · MPII valPCKh@0.5 / training cost: 89.7 -> 90.0, 66 -> 47Improved

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