The purpose of this project is to teach a neural network how to reconstruct colored images given only the grayscale images.
Here are links to some of the results I obtained with varying architectures
- Vanilla Autoencoder
- Vanilla Variational Autoencoder No-Skip
- Vanilla Variational Autoencoder UNET
- Vanilla Variational Autoencoder RESNET
Some cool results I've obtained so far from the UNET:
Image data was obtained from here: https://www.kaggle.com/datasets/theblackmamba31/landscape-image-colorization