• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site

CNN-based Post-Processing of Synthetic Objects for Image Augmentation

Student: Kalinov Aleksei

Supervisor: Anton Konushin

Faculty: Faculty of Computer Science

Educational Programme: Applied Mathematics and Information Science (Bachelor)

Final Grade: 8

Year of Graduation: 2019

Image augmentation is an essential method for increasing quality of computer vision models. Basic transformations of existing images allow to produce additional samples and promote better generalization. In some scenarios it is possible to create synthetic images with the aid of computer graphics and use them as additional data points. However, most of the synthetic examples differ from their real counterparts. This paper investigates an application of generative adversarial networks to bridge the gap between synthetic and real images as a part of image augmentation process. To validate hypothesis, experiments are performed on a highly imbalanced data set of traffic signs. Different data augmentation schemes are discussed and compared to the baseline mode. The results of the experiments are presented and the method of image post-processing that allows to increase the quality of machine learning models is outlined. In conclusion, further steps of scientific development in this field are discussed.

Full text (added May 19, 2019)

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses