Skip to content

Latest commit

 

History

History
58 lines (28 loc) · 1.01 KB

README.md

File metadata and controls

58 lines (28 loc) · 1.01 KB
Raw

DCGNA Reimplementation in TensorFLow

Link of the original paper: https://arxiv.org/abs/1511.06434

gan

Code

  1. src/DCGAN.py is the main implementation

  2. src/ops.py consists of different wrapped operations for TensorFlow

  3. main.py starts the training on MNIST dataset (for now it only trains on MNIST. Other dataset will be implemented in the future)

MINIST Results

Generator's Loss:

g_loss

Discriminator's Loss:

d_loss

Generated Numbers:

mnist_generated

CelebA Dataset

Generator's Loss:

celeba_g_loss

Discriminator's loss:

celeba_d_loss

CelebA Dataset(downloaded from http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html):

celeba_data

Generated Faces

fake_celeba

Reference

https://github.com/carpedm20/DCGAN-tensorflow

https://github.com/yihui-he/GAN-MNIST/blob/master/README.md