VAE on VizdoomTakeCover
Convolutional Variational Autoencoder on VizdoomTakeCover
In this repository I implemented a Convolutional Variational Autoencoder on frames acquired from the VizDoom environment.
Before getting started, you need to install vizdoomgym: https://github.com/simontudo/vizdoomgym
My work is based on: https://worldmodels.github.io/
Real Frame
Reconstructed Frame from my VAE
In future I will add MDN-RNN module.