search

lukemelas / unsupervised-image-segmentation

[ICLR 2022] Finding an Unsupervised Image Segmenter in each of your Deep Generative Models
77 stars 11 forks source link

A question about the pretrain of GAN #2

Closed minilionChang closed 3 years ago

minilionChang commented 3 years ago

Hello, Do you have pretraining the GAN on the CUB, Flowers datasets when you want to train a segmenter for CUB, Flowers datasets?

lukemelas commented 3 years ago

Hello,

If I understand your question correctly, you're asking if it's necessary to pretrain a GAN on Flowers/CUB in order to segment Flowers/CUB. The answer is no, you don't :) . You can use an ImageNet-pretrained GAN and simply segment images from Flowers/CUB ("zero-shot") with the same model.

minilionChang commented 3 years ago

Hello,

If I understand your question correctly, you're asking if it's necessary to pretrain a GAN on Flowers/CUB in order to segment Flowers/CUB. The answer is no, you don't :) . You can use an ImageNet-pretrained GAN and simply segment images from Flowers/CUB ("zero-shot") with the same model.

THanks for your reply!

lukemelas commented 3 years ago

Of course! Happy to help.

On Sat, May 29, 2021 at 2:38 AM minilion @.***> wrote:

Hello,

If I understand your question correctly, you're asking if it's necessary to pretrain a GAN on Flowers/CUB in order to segment Flowers/CUB. The answer is no, you don't :) . You can use an ImageNet-pretrained GAN and simply segment images from Flowers/CUB ("zero-shot") with the same model.

THanks for your reply!

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/lukemelas/unsupervised-image-segmentation/issues/2#issuecomment-850746821, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADFQ4MG62M5DVZQVYNR4T43TQBAS7ANCNFSM45SDAMDA .