search

LeeDoYup / AnoGAN-tf

Unofficial Tensorflow Implementation of AnoGAN (Anomaly GAN)
MIT License
273 stars 90 forks source link

MNIST #28

Open lzzlxxlsz opened 5 years ago

LeeDoYup commented 5 years ago

What it means?

lzzlxxlsz commented 5 years ago

sorry，I would like to ask you if your generated images are similar to the test images using the celebA dataset.I find that my result is a big difference between the test image and generated image.And when I use the MNIST dataset ,the result is all "0" images,but it is not similar with the test result.Thanks for reply !

LeeDoYup commented 5 years ago

If you use MNIST, concatenating the label into input (conditional GAN) can be a solution. I checked that the final image of celebA is fine.

lzzlxxlsz commented 5 years ago

Would you mind share your test picture and the final generated picture?What's more,I would appreciate it if you share your checkpoint file. I run your program code and find that the difference between the test picture and the final generated picture is very big, so the final reconstruction image is not good.