Abstract
| Original language | English |
|---|---|
| Title of host publication | Proceedings of European Conference on Computer Vision (ECCV) |
| Subtitle of host publication | Computer Vision - ECCV 2018 |
| Editors | Vittorio Ferrari, Martial Hebert, Cristian Sminchisescu, Yair Weiss |
| Publisher | Springer Verlag |
| Pages | 662-678 |
| Number of pages | 17 |
| Volume | 11216 |
| ISBN (Electronic) | 978-3-030-01258-8 |
| ISBN (Print) | 978-3-030-01257-1 |
| DOIs | |
| Publication status | E-pub ahead of print - 6 Oct 2018 |
Publication series
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Volume | 11216 LNCS |
| ISSN (Print) | 0302-9743 |
| ISSN (Electronic) | 1611-3349 |
Keywords
- Generative adversarial networks
- Image translation
ASJC Scopus subject areas
- Theoretical Computer Science
- Computer Science(all)
Cite this
Improving Shape Deformation in Unsupervised Image-to-Image Translation. / Gokaslan, Aaron ; Ramanujan, Vivek; Ritchie, Daniel; Kim, Kwang In; Tompkin, James.
Proceedings of European Conference on Computer Vision (ECCV): Computer Vision - ECCV 2018. ed. / Vittorio Ferrari; Martial Hebert; Cristian Sminchisescu; Yair Weiss. Vol. 11216 Springer Verlag, 2018. p. 662-678 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11216 LNCS).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Improving Shape Deformation in Unsupervised Image-to-Image Translation
AU - Gokaslan, Aaron
AU - Ramanujan, Vivek
AU - Ritchie, Daniel
AU - Kim, Kwang In
AU - Tompkin, James
PY - 2018/10/6
Y1 - 2018/10/6
N2 - Unsupervised image-to-image translation techniques are able to map local texture between two domains, but they are typically unsuccessful when the domains require larger shape change. Inspired by semantic segmentation, we introduce a discriminator with dilated convolutions that is able to use information from across the entire image to train a more context-aware generator. This is coupled with a multi-scale perceptual loss that is better able to represent error in the underlying shape of objects. We demonstrate that this design is more capable of representing shape deformation in a challenging toy dataset, plus in complex mappings with significant dataset variation between humans, dolls, and anime faces, and between cats and dogs.
AB - Unsupervised image-to-image translation techniques are able to map local texture between two domains, but they are typically unsuccessful when the domains require larger shape change. Inspired by semantic segmentation, we introduce a discriminator with dilated convolutions that is able to use information from across the entire image to train a more context-aware generator. This is coupled with a multi-scale perceptual loss that is better able to represent error in the underlying shape of objects. We demonstrate that this design is more capable of representing shape deformation in a challenging toy dataset, plus in complex mappings with significant dataset variation between humans, dolls, and anime faces, and between cats and dogs.
KW - Generative adversarial networks
KW - Image translation
UR - http://www.scopus.com/inward/record.url?scp=85055128063&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-01258-8_40
DO - 10.1007/978-3-030-01258-8_40
M3 - Conference contribution
SN - 978-3-030-01257-1
VL - 11216
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 662
EP - 678
BT - Proceedings of European Conference on Computer Vision (ECCV)
A2 - Ferrari, Vittorio
A2 - Hebert, Martial
A2 - Sminchisescu, Cristian
A2 - Weiss, Yair
PB - Springer Verlag
ER -