Thesis Details

Příprava trénovacích dat pomocí generativních neuronových sítí

Bachelor's Thesis Student: Ševčík Pavel Academic Year: 2019/2020 Supervisor: Hradiš Michal, Ing., Ph.D.
English title
Generating training data with neural networks
Language
Czech
Abstract

The aim of this thesis was to prepare a training data set for traffic sign detection using generative neural networks. The solution uses a modified U-Net architecture and several experiments with the application of styles using AdaIN layers as in the StyleGAN model have been conducted. By extending the real GTSDB data set with the generated images, mean average precision of 80.36 % has been achieved, which yields an improvement of 19.27 % compared to the mean average precision of the detection model trained on real data only.

Keywords

Generating training data, traffic sign detection, generative neural networks, GAN, SSD, Pix2Pix, AdaIN

Department
Degree Programme
Information Technology
Files
Status
defended, grade A
Date
8 July 2020
Reviewer
Committee
Čadík Martin, doc. Ing., Ph.D. (DCGM FIT BUT), předseda
Bařina David, Ing., Ph.D. (DCGM FIT BUT), člen
Burget Radek, doc. Ing., Ph.D. (DIFS FIT BUT), člen
Holík Lukáš, doc. Mgr., Ph.D. (DITS FIT BUT), člen
Jaroš Jiří, doc. Ing., Ph.D. (DCSY FIT BUT), člen
Citation
ŠEVČÍK, Pavel. Příprava trénovacích dat pomocí generativních neuronových sítí. Brno, 2020. Bachelor's Thesis. Brno University of Technology, Faculty of Information Technology. 2020-07-08. Supervised by Hradiš Michal. Available from: https://www.fit.vut.cz/study/thesis/20903/
BibTeX
@bachelorsthesis{FITBT20903,
    author = "Pavel \v{S}ev\v{c}\'{i}k",
    type = "Bachelor's thesis",
    title = "P\v{r}\'{i}prava tr\'{e}novac\'{i}ch dat pomoc\'{i} generativn\'{i}ch neuronov\'{y}ch s\'{i}t\'{i}",
    school = "Brno University of Technology, Faculty of Information Technology",
    year = 2020,
    location = "Brno, CZ",
    language = "czech",
    url = "https://www.fit.vut.cz/study/thesis/20903/"
}
Back to top