Van Huy comments

Results 36 comments of


Van Huy

What is the maximum number of iterations?

Yes! 👍

Inference runs error

It said that `NotFoundError: input_sample.jpg`. So you need to provide a input image with name `input_sample.jpg`. Of course, you can change the name as you want.

CycleGAN convergence

GAN convergence is a very difficult problem in practice. At the moment, just look at TensorBoard and stop training whenever you "feel" it enough. For this task, maybe 150k steps...

Is there anywhere different from the original CycleGAN code?

Actually that one is the same as the original code. Ironically, the original code is different from the paper at that part. I was trying to imitate the original code...

Hi, I trained your model but result not very good

Thank you for your information. I've just realized that my implementation of PatchGAN was incorrect and I'm working on it. Regards,

Hi, I trained your model but result not very good

I've removed the wrong part of PatchGAN and now the results should be better. But it takes time to confirm.

Hi, I trained your model but result not very good

Results on training set after 37 epochs. * apple2orange: * orange2apple:

Hi, I trained your model but result not very good

I tried on horse2zebra dataset but the results are still blurry. Maybe it is more difficult to translate horse2zebra than orange2apple. Did you train horse2zebra with the original pytorch version?

Hi, I trained your model but result not very good

I fixed another issue and retrained horse2zebra. The results are quite promising. A screenshot after 87 epochs:

Hi, I trained your model but result not very good

@TX2012LH just 1 line but critical https://github.com/vanhuyz/CycleGAN-TensorFlow/commit/960d1b085459856213d653d0f4e5a1c42dc1f0cf. G and F of cycle loss were reversed :flushed: