用于促进未经配对的图像到图像转换的隐式配对 " /> 用于促进未经配对的图像到图像转换的隐式配对 " /> <p class="MsoNormal" align="left" style="text-align:left;"> 用于促进未经配对的图像到图像转换的隐式配对
Please wait a minute...
Vis Inf
Vis Inf  2020, Vol. 4 Issue (4): 50-58    DOI: 10.1016/j.visinf.2020.10.001
论文     

用于促进未经配对的图像到图像转换的隐式配对

Yiftach Ginger, Dov Danon, Hadar Averbuch-Elor, Daniel Cohen-Or
Tel Aviv University, Israel
Implicit pairs for boosting unpaired image-to-image translation
Yiftach Ginger, Dov Danon, Hadar Averbuch-Elor, Daniel Cohen-Or
Tel Aviv University, Israel
 全文: PDF 
摘要： 在图像到图像的转换中，我们的目标是构建从一幅图像域到另一幅图像域的映射。在监督学习的情形中，该映射是通过已配对的样本来学习的。但是，收集大量的配对图像要么成本太高，要么不可能。因此，近年来更多的注意力聚焦在如何从未经配对的图像中来学习映射的技术。 在我们的工作中，我们将隐式配对注入未经配对的图像集合中，来加强两个图像域之间的映射，提高它们的分布的兼容性，从而使无监督技术在多项测量中的性能提升高达12％。

我们通过使用伪配对（即近似于真实配对的一些样本对），进一步展示了隐式配对的能力。我们演示了这些近似的隐式配对样本对图像到图像转换的作用，这些伪配对可能在一个方向上取近似，而在另一个方向上不加改动。我们进一步展示，在未经配对的图像转换中，伪配对作为隐式配对使用比在配对图像转换中直接显式使用更为有效。

关键词： 生成对抗网络;  图像到图像的转换;  数据扩充;  合成样本    
Abstract: In image-to-image translation the goal is to learn a mapping from one image domain to another. In the case of supervised approaches the mapping is learned from paired samples. However, collecting large sets of image pairs is often either prohibitively expensive or not possible. As a result, in recent years more attention has been given to techniques that learn the mapping from unpaired sets. In our work, we show that injecting implicit pairs into unpaired sets strengthens the mapping between the two domains, improves the compatibility of their distributions, and leads to performance boosting of unsupervised techniques by up to 12% across several measurements. The competence of the implicit pairs is further displayed with the use of pseudo-pairs, i.e., paired samples which only approximate a real pair. We demonstrate the effect of the approximated implicit samples on image-to-image translation problems, where such pseudo-pairs may be synthesized in one direction, but not in the other. We further show that pseudo-pairs are significantly more effective as implicit pairs in an unpaired setting, than directly using them explicitly in a paired setting.

Key words: Generative adversarial networks    Image-to-image translation    Data augmentation    Synthetic samples
出版日期: 2020-12-01
通讯作者: Yiftach Ginger     E-mail: iftachg@mail.tau.ac.il
服务　　
把本文推荐给朋友 用于促进未经配对的图像到图像转换的隐式配对 ”的文章，特向您推荐。请打开下面的网址：http://www.zjujournals.com/vi/CN/abstract/abstract41633.shtml" name="neirong"> 用于促进未经配对的图像到图像转换的隐式配对 ">
加入引用管理器
E-mail Alert
RSS
作者相关文章　　
Yiftach Ginger
Dov Danon
Hadar Averbuch-Elor
Daniel Cohen-Or

引用本文:

Yiftach Ginger, Dov Danon, Hadar Averbuch-Elor, Daniel Cohen-Or. Implicit pairs for boosting unpaired image-to-image translation . Vis Inf, 2020, 4(4): 50-58.

链接本文:

http://www.zjujournals.com/vi/CN/10.1016/j.visinf.2020.10.001    或    http://www.zjujournals.com/vi/CN/Y2020/V4/I4/50

No related articles found!