File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Conference Paper: Learning to Manipulate Individual Objects in an Image

TitleLearning to Manipulate Individual Objects in an Image
Authors
Issue Date2020
Citation
Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2020, p. 6557-6566 How to Cite?
AbstractWe describe a method to train a generative model with latent factors that are (approximately) independent and localized. This means that perturbing the latent variables affects only local regions of the synthesized image, corresponding to objects. Unlike other unsupervised generative models, ours enables object-centric manipulation, without requiring object-level annotations, or any form of annotation for that matter. The key to our method is the combination of spatial disentanglement, enforced by a Contextual Information Separation loss, and perceptual cycle-consistency, enforced by a loss that penalizes changes in the image partition in response to perturbations of the latent factors. We test our method's ability to allow independent control of spatial and semantic factors of variability on existing datasets and also introduce two new ones that highlight the limitations of current methods.
Persistent Identifierhttp://hdl.handle.net/10722/325497
ISSN
2023 SCImago Journal Rankings: 10.331
ISI Accession Number ID

 

DC FieldValueLanguage
dc.contributor.authorYang, Yanchao-
dc.contributor.authorChen, Yutong-
dc.contributor.authorSoatto, Stefano-
dc.date.accessioned2023-02-27T07:33:46Z-
dc.date.available2023-02-27T07:33:46Z-
dc.date.issued2020-
dc.identifier.citationProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2020, p. 6557-6566-
dc.identifier.issn1063-6919-
dc.identifier.urihttp://hdl.handle.net/10722/325497-
dc.description.abstractWe describe a method to train a generative model with latent factors that are (approximately) independent and localized. This means that perturbing the latent variables affects only local regions of the synthesized image, corresponding to objects. Unlike other unsupervised generative models, ours enables object-centric manipulation, without requiring object-level annotations, or any form of annotation for that matter. The key to our method is the combination of spatial disentanglement, enforced by a Contextual Information Separation loss, and perceptual cycle-consistency, enforced by a loss that penalizes changes in the image partition in response to perturbations of the latent factors. We test our method's ability to allow independent control of spatial and semantic factors of variability on existing datasets and also introduce two new ones that highlight the limitations of current methods.-
dc.languageeng-
dc.relation.ispartofProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition-
dc.titleLearning to Manipulate Individual Objects in an Image-
dc.typeConference_Paper-
dc.description.naturelink_to_subscribed_fulltext-
dc.identifier.doi10.1109/CVPR42600.2020.00659-
dc.identifier.scopuseid_2-s2.0-85094842526-
dc.identifier.spage6557-
dc.identifier.epage6566-
dc.identifier.isiWOS:000620679506083-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats