Skip to content

Collection of python scripts running VAE models using Keras and Tensorflow as backend

Notifications You must be signed in to change notification settings

anktplwl91/vae_models

Repository files navigation

vae_models

Collection of python scripts running VAE models using Keras and Tensorflow as backend

This repository is a collection of VAE-scripts for image classification and generation and text classification and generation (upcoming)

  1. fashion_autoencoder.py has a python script that builds a VAE model using Keras for Fashion-MNIST dataset. Following are some images that were generated by the network.

Representaion of images in test-set in two-dimensional latent space Representaion of original images in two-dimensional space

Some reconstructed images by Decoder network gen_images

A set of original input images and corresponding reconstructed images Input Images Output Images

There are three different convolutional layers in the Encoder network, each of them activates the input image at different levels.

orig_activations

Similar is the case for the Decoder network. gen_activations

About

Collection of python scripts running VAE models using Keras and Tensorflow as backend

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages