What is a GAN?
What is a GAN?
Generative Adversarial Networks are deep learning models that consist of two neural networks a generator and a discriminator. These models are primarily used to generate images from noise.
Generative Adversarial Networks are deep learning models that consist of two neural networks a generator and a discriminator. These models are primarily used to generate images from noise.
Generating fake images allow for datasets to increase which will allow for more availability of data to train deep learning models.
Much of the research and development in image synthesis and GANs in general deal with training models with human or scene images. There isn't much experimentation in animation or art when stacked with the number of research done with 'real life' images. It would also be a fun topic to learn how GANs work.
Absolutely, being able to generate anime can be further improved on to create anime from small drawings. Allowing content creators or artists to take full advantage of this.
This was a simple DCGAN model which quickly since the generator was able to fool the discriminator quickly. So in order to improve the results we need to lock the generator and train the discriminator to further improve this specific model.