/DCGAN-on-Anime-Faces

The respository contains the code for the Deep Convolutional Generative Adversarial Network (DCGAN) trained on the anime faces dataset.

Primary LanguagePython

DCGAN-on-Anime-Faces

The respository contains the code for the Deep Convolutional Generative Adversarial Network (DCGAN) trained on the anime faces dataset.

Blog post: https://idiotdeveloper.com/dcgan-implementing-deep-convolutional-generative-adversarial-network-in-tensorflow/
Video explaination: https://youtu.be/rJBp2L-JM_g

There are two files:

  • gan.py: It is used to train the DCGAN on the anime faces dataset.
  • test.py: This file loads the saved generator model and generates some fake samples and save them with the name fake.png

Dataset

Download the dataset and put all the images in the data folder.
Link: https://www.kaggle.com/soumikrakshit/anime-faces

Training samples

Image at epoch 1
Image at epoch 1
Image at epoch 1
Image at epoch 60

Fake samples generated after training

Fake sample