Generator Model producing only black images. Cannot find problem in Neural Network

1k views Asked by At

I have been working on making a python generator that generates anime characters using a DeepGAN; however, I have finished the model but it still produces just black images. I am not sure what is the problem in my code though I suspect the way I am saving my model. I would like some help or just another set of eyes to help me find the error and solve the problem.

I posted below my notebook. I would appreciate if you would check out the notebook where all my code and output is posted. (problem is in the end last cells of the notebook, and the notebook is pretty short)

https://github.com/SarthakJaingit/GenerateAnimecharacters/blob/master/notebook683adad691%20(1).ipynb Please look at the end for the snippet of code shown below.

Thanks a lot for the help, Sarthak Jain

The code which produces the black images is:

image = G(vector)
plt.imshow(ima.reshape(64, 64, 3))
#output: Generator (G) produces black image out of vector

D_out = D(image)
print(real_loss(D_out))
# Discriminator (D) Takes in the black image and says it looks real 


0

There are 0 answers