jacobgil / keras-dcgan

Keras implementation of Deep Convolutional Generative Adversarial Networks

Why did not use Conv2DTranspose rather than Conv2D in the generator

lestel opened this issue 6 years ago · comments

lestel commented 6 years ago

the related paper all say they use Conv2DTranspose

Nagabhushan S N commented 5 years ago

Yes. That is one of their key contributions.
They say to

use strided convolutions instead of max-pooling layer
use cond2dTranspose instead of upsampling
use batch normalization layers
use relu activations for intermediate layers

@jacobgil You've implemented none of those. So, this is not DC GAN right?