Generative adversarial networks are a type of artificial intelligence algorithms used in unsupervised machine learning, implemented by a system of two neural networks competing against each other in a zero-sum game framework. I have used Deep Convolutional Generative Adversarial Networks to generate Pokemon images.
- Tensorflow
- Keras
- Pillow install missing dependencies using pip
Run python Main.py <Epochs for training> <Batch size for training> <Batch size for generation> in the terminal. GAN's training is quite complex, training in a GPU environment is highly recommended for better results.
The dataset is available here. Move the additional dataset to logos folder.
This work is learned and is highly based on Youtube Tutorial by Siraj Raval.