AI Paper Atlas

Research map / 3D vision / Human pose estimation

Few-Shot Human Motion Prediction via Meta-Learning

Liang-Yan Gui, Yu-Xiong Wang, Deva Ramanan, Jose M. F. Moura

ECCV 2018 · 3D vision › Human pose estimation · approach: Human motion prediction · 126 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 meta-learned generic initialization (MAML) to adaptation-only model (ablation)PAML · H3.6M Walking 80msmean angle error: 0.39 -> 0.35Improved
add adaptation network H to init-only model (ablation)PAML · H3.6M Smoking 80msmean angle error: 0.53 -> 0.39Improved
add adaptation network H to init-only model (ablation)PAML · H3.6M Discussion 80msmean angle error: 0.54 -> 0.41Improved
add meta-learned initialization to adaptation-only model (ablation)PAML · H3.6M Eating 80msmean angle error: 0.50 -> 0.36Improved
meta-learning with PAML instead of fine-tuning transferresidual sup. · H3.6M Walking 80msmean angle error: 0.44 -> 0.35Improved
PAML instead of multi-task learningresidual sup. · H3.6M Smoking 80msmean angle error: 0.71 -> 0.39Improved
action-agnostic instead of action-specific training from scratchresidual sup. · H3.6M Walking 80msmean angle error: 1.90 -> 1.78Improved
use 3-layer Leaky ReLU H instead of affine 1-layer H (ablation)PAML · H3.6M Walking 80msmean angle error: 0.39 -> 0.35Improved
use Leaky ReLU instead of ReLU in 3-layer H (ablation)PAML · H3.6M Walking 80msmean angle error: 0.37 -> 0.35Improved
go from 3 to 4 layers in H (ablation)PAML · H3.6M Walking 80msmean angle error: 0.35 -> 0.36Hurt
PAML instead of MAML in few-shot classificationPAML · mini-ImageNet 5-way 1-shotaccuracy: 48.70% -> 53.26%Improved
PAML instead of MAML in few-shot classificationPAML · mini-ImageNet 5-way 5-shotaccuracy: 63.11% -> 68.19%Improved

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