Skip to main navigation Skip to search Skip to main content

Deforming autoencoders: Unsupervised disentangling of shape and appearance

  • Zhixin Shu
  • , Mihir Sahasrabudhe
  • , Rıza Alp Güler
  • , Dimitris Samaras
  • , Nikos Paragios
  • , Iasonas Kokkinos
  • Stony Brook University
  • Université Paris-Saclay
  • Institut national de recherche en informatique et en automatique
  • TheraPanacea
  • University College London
  • Facebook AI Research

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

35 Scopus citations

Abstract

In this work we introduce Deforming Autoencoders, a generative model for images that disentangles shape from appearance in an unsupervised manner. As in the deformable template paradigm, shape is represented as a deformation between a canonical coordinate system (‘template’) and an observed image, while appearance is modeled in deformation-invariant, template coordinates. We introduce novel techniques that allow this approach to be deployed in the setting of autoencoders and show that this method can be used for unsupervised group-wise image alignment. We show experiments with expression morphing in humans, hands, and digits, face manipulation, such as shape and appearance interpolation, as well as unsupervised landmark localization. We also achieve a more powerful form of unsupervised disentangling in template coordinates, that successfully decomposes face images into shading and albedo, allowing us to further manipulate face images.

Original languageEnglish
Title of host publicationComputer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings
EditorsMartial Hebert, Vittorio Ferrari, Cristian Sminchisescu, Yair Weiss
PublisherSpringer Verlag
Pages664-680
Number of pages17
ISBN (Print)9783030012489
DOIs
StatePublished - 2018
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)
Volume11214 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference15th European Conference on Computer Vision, ECCV 2018
Country/TerritoryGermany
CityMunich
Period09/8/1809/14/18

Fingerprint

Dive into the research topics of 'Deforming autoencoders: Unsupervised disentangling of shape and appearance'. Together they form a unique fingerprint.

Cite this