Publication: Photo Enhancement On Mobile Devices Using Deep Neural Networks
Loading...
Identifiers
Publication date
2019
Defense date
2019-09
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In recent years, the return of the usage of Artificial Neural Networks has lead to the
greatest improvements in the field of Artificial Intelligence, due to the huge diversity
of different applications that deep learning models has in a large variety of research
fields, and also the evolution of information processing systems capacity. This thesis
aims to study which deep neural networks models are most suitable for photo
enhancement, to generate images with certain desired characteristics.
Model selection has been done by comparing the both supervised, Convolutional
Neural Networks, and unsupervised models, Generative Adversarial Networks. It has
been demonstrated that Generative Adversarial Networks have great potential by
showing results that compete with the state of the art. The chosen model is a
Generative Adversarial model which outperforms the rest in terms of a combination
of enhancement quality and time taken in the process. Moreover, since the model is
compatible with mobile devices it has been integrated and evaluated in a BQ
smartphone, to proof its viability on mobile devices.
Description
Keywords
Mejora de imágenes, Redes neuronales, Aplicaciones móviles