tgeorgy / mgan Goto Github PK

Hi, tgeorgy.

I really appreciate your work.
And i wonder whether u wrote any paper about this project?

Valentina

Hi, great work!

How to apply this multiple faces? i mean the images which includes more than 1 person

regards

Hi, thank you for sharing this code, but when I executing 'sh init_data.sh', I got following error, may you provide a solution for me?

extracting: img_align_celeba/202594.jpg
extracting: img_align_celeba/202595.jpg
extracting: img_align_celeba/202596.jpg
extracting: img_align_celeba/202597.jpg
extracting: img_align_celeba/202598.jpg
extracting: img_align_celeba/202599.jpg
Traceback (most recent call last):
File "crop_images.py", line 29, in
os.rename(ipath+fname, 'data/celeba/pos/'+fname)
OSError: [Errno 2] No such file or directory

Hi, tgeorgy.

I'v just started learning GAN and used your code for my first try. Unluckily I met a problem which I can't solve. At line 113 an error existed, where the code is:

loss_P2N_gan = criterion_gan(fake_neg_lbl, Variable(real_lbl)).

And the traceback is :

RuntimeError: input and target shapes do not match: input [1 x 1 x 1 x 1], target [1 x 1]

It seems that the dimensions of the scalars differ from each other. Could you tell me how to solve the problem? Or what should I learn to handle this? I ran the code using python 3.6, so I changed the 'itertools.izip' to 'zip', does this affect the programme?

Best regards,
Hondahmd

How long have you train your model?

can u provide a test.py.
thanks

Hi~ Seeing from the graph you provided , the generated image containing mask will go through discriminator , but in your code it seems you didn't do that?