TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

SinGAN: Learning a Generative Model from a Single Natural Image

77 pointsby groarover 5 years ago

5 comments

jeeceebeesover 5 years ago
I think the approach is really cool but the processing time required is too much for this to be very useful at the moment.<p>On a 1080 Ti it takes 45-90 minutes to train networks for the various tasks on 256px images (depending on some quality parameters and which task). Each task also requires training individually so if you&#x27;d like to try them all for a given image you&#x27;ll need to train 6 times.<p>Also the pyramid of GANs approach is very memory hungry. I was only able to get up to 724px images with 11 GB of VRAM. This was also only possible with a higher scale factor (sparser pyramid) which sacrifices a lot of quality and is incredibly noticeable at larger image sizes. I only tried for larger sizes with the animation task though, perhaps there is a way to combine the super resolution and animation task and achieve better results. Training on larger sizes was taking upwards of 6-8 hours.<p>All of this was tested with the official repo[1] about a month ago.<p>[1] <a href="https:&#x2F;&#x2F;github.com&#x2F;tamarott&#x2F;SinGAN" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;tamarott&#x2F;SinGAN</a>
评论 #21638781 未加载
maffydubover 5 years ago
Not read the paper in detail yet, but it reminds me of Deep Image Prior (<a href="https:&#x2F;&#x2F;sites.skoltech.ru&#x2F;app&#x2F;data&#x2F;uploads&#x2F;sites&#x2F;25&#x2F;2018&#x2F;04&#x2F;deep_image_prior.pdf" rel="nofollow">https:&#x2F;&#x2F;sites.skoltech.ru&#x2F;app&#x2F;data&#x2F;uploads&#x2F;sites&#x2F;25&#x2F;2018&#x2F;04&#x2F;...</a>).
评论 #21637840 未加载
kidintechover 5 years ago
This looks fantastic and gets me excited about where the field is going, despite the performance issues.
ticktocktenover 5 years ago
Having spent some time in trying to do style transfers, this looks very promising.<p>The harmonization aspect of the paper actually makes it very useful. There certainly are cases where you want to introduce an image component as an overlay and want the style to integrate.<p>Really cool stuff, and with code!
sharemywinover 5 years ago
where there any images with people?
评论 #21639123 未加载