Image impainting using deep neural networks

Nowadays, virtual markets are increasingly available and seek to connect shoppers with products. Due to the high turnover of products in a physical market, it is very likely to find relevant differences between products in the physical and digital markets. This paper proposes the use of image inpain...

Full description

Bibliographic Details
Author: Zurita Montes De Oca, Erika Anabel
Format: master thesis
Publication Date:2023
Country:España
Institution:Universitat Politècnica de Catalunya (UPC)
Repository:UPCommons. Portal del coneixement obert de la UPC
Language:English
OAI Identifier:oai:upcommons.upc.edu:2117/388362
Online Access:https://hdl.handle.net/2117/388362
Access Level:Open access
Keyword:Neural networks (Computer science)
Deep learning
Image processing
Image impainting
Neural networks
Machine learning
Self-supervised neural networks
Generative adversarial networks
Xarxes neuronals (Informàtica)
Aprenentatge profund
Àrees temàtiques de la UPC::Enginyeria de la telecomunicació
Description
Summary:Nowadays, virtual markets are increasingly available and seek to connect shoppers with products. Due to the high turnover of products in a physical market, it is very likely to find relevant differences between products in the physical and digital markets. This paper proposes the use of image inpainting using deep neural networks to solve this problem. It is proposed to use the approach performed by [1] based on generative adversarial networks as they are one of the most inventive and promising architectures. Through the experiments performed, it has been possible to prove that using this method it is possible to train models that produce realistic terminations of products that have been eliminated or that are to be replaced. We have also made a comparison with another interesting approach that had shown good results in the task of content generation in arbitrary zones.