I trained a Generative Adversarial Network (GAN) to generate images of new Pokémon. I created an API using Flask and a frontend to interact with the trained model.
GAN Python Flask
Machine learning is something really interesting I wanted to try. To do so, I decided to implement a Generative Adversarial Network (GAN) using tensorflow and Python based on documentation I found online. After building the model, I trained it using Google Colab with a extended Pokemon images dataset. The results I had were far from perfect but I couldn’t train my model longer or with higher resolution images because of technical limitations. In the end, I decided it was good enough to show it off on a website I created.
The website is very simple and uses a Flask backend that loads the model with the trained weights and use it to generate images.
The Pokemon generator website can be found here: Pokemon GAN