hiteshnitetc

Should we have to use as many epoch as we have no. of variable size images?

Waiting for response

Have to got solution

You Got solution?

You can save your own data into .npy file into google drive, then use np.load, to call data.

> > > Hello, i also want to use the other datasets (like: ck+) , but i failed to complete it(DCGAN). > i couldn't find the way to address this...

Please go through following lines: d_loss_real = self.discriminator.train_on_batch(imgs, [valid, labels], class_weight=[cw1, cw2]) d_loss_fake = self.discriminator.train_on_batch(gen_imgs, [fake, fake_labels], class_weight=[cw1, cw2]) In first training, valid labels are used, in second, fake labels...

In def sample_images: self.discriminator.save("location/dweights_%d.hdf5" %epoch) (for discriminator) self.generator.save("location/gweights_%d.hdf5" %epoch) (for generator) you can also save model in JSON, and use save_weights for saving weights.

Just use load_model, and then train. On Sat, Jul 11, 2020 at 8:29 PM KamiKazeFei wrote: > @hiteshnitetc > after I save the weight and model, how to do can...

Have you got answer? I have same question regarding SGAN? I thing it is label classification accuracy (not sure).