Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Generation of Artificial Training Data for Deep Learning

Wessman, David LU and Andersson, Pontus (2018) In Master’s Theses in Mathematical Sciences FMAM05 20181
Mathematics (Faculty of Engineering)
Abstract
Can artificial training data be used for deep learning applications in computer vision? We investigated this by building a framework that uses computer graphics to generate large quantities of images portraying humans. Generative Adversarial Networks (GANs) were used to bridge the gap in appearance between the generated and real images. A dataset’s potential was estimated by first using it to train a person re-identification model, and then evaluating the model on real images. Results showed that datasets that had been put through a GAN had higher potential than those which had not. We were not able to replace real images with artificial ones, but our results showed promise for further work in substituting and complementing real images.
Please use this url to cite or link to this publication:
author
Wessman, David LU and Andersson, Pontus
supervisor
organization
course
FMAM05 20181
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Artificial training data, deep learning, generative adversarial networks, large scale image generation, person re-identification.
publication/series
Master’s Theses in Mathematical Sciences
report number
LUTFMA-3359-2018
ISSN
1404-6342
other publication id
2018:E58
language
English
id
8956843
date added to LUP
2018-09-17 14:28:28
date last changed
2018-09-17 14:28:28
@misc{8956843,
  abstract     = {{Can artificial training data be used for deep learning applications in computer vision? We investigated this by building a framework that uses computer graphics to generate large quantities of images portraying humans. Generative Adversarial Networks (GANs) were used to bridge the gap in appearance between the generated and real images. A dataset’s potential was estimated by first using it to train a person re-identification model, and then evaluating the model on real images. Results showed that datasets that had been put through a GAN had higher potential than those which had not. We were not able to replace real images with artificial ones, but our results showed promise for further work in substituting and complementing real images.}},
  author       = {{Wessman, David and Andersson, Pontus}},
  issn         = {{1404-6342}},
  language     = {{eng}},
  note         = {{Student Paper}},
  series       = {{Master’s Theses in Mathematical Sciences}},
  title        = {{Generation of Artificial Training Data for Deep Learning}},
  year         = {{2018}},
}