Unsupervised Image-to-Image Translation via Pre-trained StyleGAN2 Network
release_26x2cogfgjaaxesh3xenvnralq
by
Jialu Huang, Jing Liao, Sam Kwong
2020
Abstract
Image-to-Image (I2I) translation is a heated topic in academia, and it also
has been applied in real-world industry for tasks like image synthesis,
super-resolution, and colorization. However, traditional I2I translation
methods train data in two or more domains together. This requires lots of
computation resources. Moreover, the results are of lower quality, and they
contain many more artifacts. The training process could be unstable when the
data in different domains are not balanced, and modal collapse is more likely
to happen. We proposed a new I2I translation method that generates a new model
in the target domain via a series of model transformations on a pre-trained
StyleGAN2 model in the source domain. After that, we proposed an inversion
method to achieve the conversion between an image and its latent vector. By
feeding the latent vector into the generated model, we can perform I2I
translation between the source domain and target domain. Both qualitative and
quantitative evaluations were conducted to prove that the proposed method can
achieve outstanding performance in terms of image quality, diversity and
semantic similarity to the input and reference images compared to
state-of-the-art works.
In text/plain
format
Archived Content
There are no accessible files associated with this release. You could check other releases for this work for an accessible version.
Know of a fulltext copy of on the public web? Submit a URL and we will archive it
2010.05713v1
access all versions, variants, and formats of this works (eg, pre-prints)