Several methods were recently proposed for the task of translating images between domains without prior knowledge in the form of correspondences. The existing methods apply adversarial learning to ensure that the distribution of the mapped source domain is indistinguishable from the target domain, which suffers from known stability issues. In addition, most methods rely heavily on “cycle” relationships between the domains, which enforce a one-to-one mapping. In this work, we introduce an alternative method: Non-Adversarial Mapping (NAM), which separates the task of target domain generative modeling from the cross-domain mapping task. NAM relies on a pre-trained generative model of the target domain, and aligns each source image with an image synthesized from the target domain, while jointly optimizing the domain mapping function. It has several key advantages: higher quality and resolution image translations, simpler and more stable training and reusable target models. Extensive experiments are presented validating the advantages of our method.
|Original language||American English|
|Title of host publication||Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings|
|Editors||Vittorio Ferrari, Cristian Sminchisescu, Yair Weiss, Martial Hebert|
|Number of pages||16|
|State||Published - 2018|
|Event||15th European Conference on Computer Vision, ECCV 2018 - Munich, Germany|
Duration: 8 Sep 2018 → 14 Sep 2018
|Name||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|Conference||15th European Conference on Computer Vision, ECCV 2018|
|Period||8/09/18 → 14/09/18|
Bibliographical notePublisher Copyright:
© 2018, Springer Nature Switzerland AG.