About CcGAN

[Kanon5535]

Hello, I'm Shi from Tokyo.

I have created my program with reference to your program. However, when I ran it, I got the error shown in the photos. By the way, when I change batch_size_d to 8, my program run, but the d_loss are always 0. I tried to solved it, but I had no idea what to do, so I hope you can help me. I'm Chinese, and I can read and speak Chinese, if you feel easy to write in Chinese you can use it.

[UBCDingXin]

Hi Shi,

Thanks for your interest in our work.

As for your problem, I suggest double-checking the dimension of the output of each block in the discriminator. It seems there is a dimension mismatch problem.

Hope it can help.

[Kanon5535]

Hi Ding,

Thanks for replying. I made some changes to my program based on what you told me, and the problem has been resolved. And now I want to generate images with specific labels, for example, only the chair with 55° angle. I tried to write a test　program to load the trained model and generate the images, but it seems to fail when adding the labels. I would like to ask, how can I add the specific labels right? I apologize for the inconvenience, but I would appreciate your reply.

[UBCDingXin]

We have detailed descriptions of the improved label mechanism in our paper. You can also check our code for the detailed implementation.