Few-shot human motion prediction via meta-learning

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

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Human motion prediction, forecasting human motion in a few milliseconds conditioning on a historical 3D skeleton sequence, is a long-standing problem in computer vision and robotic vision. Existing forecasting algorithms rely on extensive annotated motion capture data and are brittle to novel actions. This paper addresses the problem of few-shot human motion prediction, in the spirit of the recent progress on few-shot learning and meta-learning. More precisely, our approach is based on the insight that having a good generalization from few examples relies on both a generic initial model and an effective strategy for adapting this model to novel tasks. To accomplish this, we propose proactive and adaptive meta-learning (PAML) that introduces a novel combination of model-agnostic meta-learning and model regression networks and unifies them into an integrated, end-to-end framework. By doing so, our meta-learner produces a generic initial model through aggregating contextual information from a variety of prediction tasks, while effectively adapting this model for use as a task-specific one by leveraging learning-to-learn knowledge about how to transform few-shot model parameters to many-shot model parameters. The resulting PAML predictor model significantly improves the prediction performance on the heavily benchmarked H3.6M dataset in the small-sample size regime.

Original languageEnglish (US)
Title of host publicationComputer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings
EditorsVittorio Ferrari, Cristian Sminchisescu, Yair Weiss, Martial Hebert
PublisherSpringer
Pages441-459
Number of pages19
ISBN (Print)9783030012366
DOIs
StatePublished - 2018
Externally publishedYes
Event15th European Conference on Computer Vision, ECCV 2018 - Munich, Germany
Duration: Sep 8 2018Sep 14 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11212 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other15th European Conference on Computer Vision, ECCV 2018
Country/TerritoryGermany
CityMunich
Period9/8/189/14/18

Keywords

  • Few-shot learning
  • Human motion prediction
  • Meta-learning

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Few-shot human motion prediction via meta-learning'. Together they form a unique fingerprint.

Cite this