Thesis Details

Využití sítí typu GAN pro zpřesňování detekce a rozpoznávání dopravních značek

Bachelor's Thesis Student: Glos Michal Academic Year: 2020/2021 Supervisor: Smrž Pavel, doc. RNDr., Ph.D.
English title
Improving Accuracy of Detection and Recognition of Traffic Signs with GANs
Language
Czech
Abstract

The goal of this thesis was to extend a dataset for traffic sign detection. The solution was based on generative neural networks PatchGAN and Wasserstein GAN of combined DenseNet and U-Net architecture. Those models were designed to synthesize real looking traffic signs from images of their norms. Model for object detection SSD, trained on synthetic data only, achieved mean average precision of 59.6 %, which is an improvement of 9.4 % over the model trained on the original data. SSD model trained on synthetic and original data combined achieved mean average precision of 80.1 %.

Keywords

SD, GAN, traffic sign detection, generative model, Pix2Pix, U-Net, Wasserstein GAN, PatchGAN.

Department
Department of Computer Graphics and Multimedia FIT BUT
Degree Programme
Information Technology
Files
Status
defended, grade B
Date
16 June 2021
Reviewer
Musil Petr, Ing., Ph.D.
Committee
Smrž Pavel, doc. RNDr., Ph.D. (DCGM FIT BUT), předseda
Burgetová Ivana, Ing., Ph.D. (DIFS FIT BUT), člen
Kreslíková Jitka, doc. RNDr., CSc. (DIFS FIT BUT), člen
Peringer Petr, Dr. Ing. (DITS FIT BUT), člen
Strnadel Josef, Ing., Ph.D. (DCSY FIT BUT), člen
Citation
GLOS, Michal. Využití sítí typu GAN pro zpřesňování detekce a rozpoznávání dopravních značek. Brno, 2021. Bachelor's Thesis. Brno University of Technology, Faculty of Information Technology. 2021-06-16. Supervised by Smrž Pavel. Available from: https://www.fit.vut.cz/study/thesis/23866/
BibTeX 
@bachelorsthesis{FITBT23866,
    author = "Michal Glos",
    type = "Bachelor's thesis",
    title = "Vyu\v{z}it\'{i} s\'{i}t\'{i} typu GAN pro zp\v{r}es\v{n}ov\'{a}n\'{i} detekce a rozpozn\'{a}v\'{a}n\'{i} dopravn\'{i}ch zna\v{c}ek",
    school = "Brno University of Technology, Faculty of Information Technology",
    year = 2021,
    location = "Brno, CZ",
    language = "czech",
    url = "https://www.fit.vut.cz/study/thesis/23866/"
}

Theses

Back to top