Yunsheng Li

The images used to train CycleGAN should be first resized to the same scale e.g. the height or width is 1024. Then due to the GPU memory limitation, a 452x452...

No, the resolution used to train both segmentation model and cyclegan is the same. For cycelgan, I just crop the images to the size 452x452. You can check my code...

The process is correct. But I do not think you can change the image resolution when CycleGAN is trained. Because resizing cannot recover the loss of resolution.

Please use the default parameters. The change I made is only --resize_or_crop, scale_width_and_crop --niter 10 --niter_decay 10, --loadSize 1024 --fineSize 452, --lambda_semantic_A 1 --lambda_semantic_B 1

For your first question, I tried both scenarios and get similar results. Thus, you can train CycleGAN without perceptual loss. For the second one, the model is pertained with ImageNet.

I use 4 gpus

You can train with less epochs. I upload the parameters I use. You can refer to it.

Yes. You are right.

The code has been uploaded to the folder named 'cyclegan'. You can download the original project for cyclegan from their website and merge the code I uploaded to it.

Sorry for the confusion. It is a bug and I just change it to deeplab. On Thu, Jan 9, 2020 at 11:50 PM Sicheng Xu wrote: > @liyunsheng13 ,hi, could...