Face autoencoder
Recreate pictures of human faces using a deep learning autoencoder
-
Date: 19/01/2020
Languages: Python, C#
Keywords: Tensorflow, Keras, Unity
Overview
This is a deep learning auto encoder created in python using the Tensorflow and Keras libaraies.
The autoencoder read in 2000 pictures of faces to be trained an then tested on. Once trained, it was able to be given a new face and recreate it. This recreation was then evaluated for it’s accuracy which was then used to update the autoencoder’s neural network.
The output face the autoencoder produced have be put into a demo in unity as the faces of npc’s.
Demo video
Video demonstrating the autoencoder, as well as the plant generator and ray-tracer, also made alongside this project.