Graphic Design with Machine Learning: Fabric PatternGeneration using a Variational Autoencoder

Bachelor Thesis (2021)
Author(s)

J.A. Wallage (TU Delft - Electrical Engineering, Mathematics and Computer Science)

Contributor(s)

J.C. van Gemert – Mentor (TU Delft - Pattern Recognition and Bioinformatics)

B. Yildiz – Mentor (TU Delft - Pattern Recognition and Bioinformatics)

Faculty
Electrical Engineering, Mathematics and Computer Science
Copyright
© 2021 Just Wallage
More Info
expand_more
Publication Year
2021
Language
English
Copyright
© 2021 Just Wallage
Graduation Date
01-07-2021
Awarding Institution
Delft University of Technology
Project
['CSE3000 Research Project']
Programme
['Computer Science and Engineering']
Faculty
Electrical Engineering, Mathematics and Computer Science
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

The goal of this research is to find out whether it is possible to use a specific type of machine learning algorithm to create new data that resembles data in a given dataset. The algorithm that is used is a variational autoencoder (VAE), this is a machine learning algorithm based around data compression and decompression. The dataset is a set of roughly 47,000 images of colorful fabric pattern designs created bya company called Vlisco. The working of a VAE is explained and implementations are discussed. Samples of generated images are shown and their quality is discussed.Eventually the results are compared to the results of a different research that focuses on the same problem but with a general adversarial network (GAN). If such algorithms work well they could potentially provide designers with new designs or inspiration quickly.

Files

License info not available
Color_generation.gif
(gif | 12.2 Mb)
License info not available
Greyscale_generation.gif
(gif | 4.59 Mb)
License info not available