search

dorarad / gansformer

Generative Adversarial Transformers
MIT License
1.32k stars 149 forks source link

data resolution #27

Closed leesky1c closed 2 years ago

leesky1c commented 2 years ago

Hi, thanks for ur repo. Does it support the rectangular images for training, for example, a 1024x512 image?

dorarad commented 2 years ago

Currently the way that I approach this care is by generating a square image (with zero padding) and cropping it at the end. I'll add automatic support in the repo and get back to you!

From: Leesky @.> Sent: Friday, October 8, 2021 4:17 AM To: dorarad/gansformer @.> Cc: Subscribed @.***> Subject: [dorarad/gansformer] data resolution (#27)

Hi, thanks for ur repo. Does it support the rectangular images for training, for example, a 1024x512 image?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/dorarad/gansformer/issues/27, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AFNR43E2CYCUYROL64S3RDLUF3HK7ANCNFSM5FTMHXRA. Triage notifications on the go with GitHub Mobile for iOShttps://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Androidhttps://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

dorarad commented 2 years ago

Hi, training on different image ratios and resolutions is now supported in both the TF and pytorch versions! You can specify them by e.g. --resolution 1024 --ratio 0.5 (which means height=512, width=1024)