Cycle route signs detection using deep learning
This article addresses the issue of detecting traffic signs signalling cycle routes. It is also necessary to read the number or text of the cycle route from the given image. These tags are kept under the identifier IS21 and have a defined, uniform design with text in the middle of the tag. The detec...
Saved in:
Main Authors: | , , , , |
---|---|
Format: | Article |
Published: |
Springer Science and Business Media Deutschland GmbH
2022
|
Subjects: | |
Online Access: | http://eprints.utm.my/id/eprint/100514/ http://dx.doi.org/10.1007/978-3-031-16014-1_8 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
id |
my.utm.100514 |
---|---|
record_format |
eprints |
spelling |
my.utm.1005142023-04-14T02:41:11Z http://eprints.utm.my/id/eprint/100514/ Cycle route signs detection using deep learning Kopecky, Lukas Dobrovolny, Michal Fuchs, Antonin Selamat, Ali Krejcar, Ondrej QA76 Computer software This article addresses the issue of detecting traffic signs signalling cycle routes. It is also necessary to read the number or text of the cycle route from the given image. These tags are kept under the identifier IS21 and have a defined, uniform design with text in the middle of the tag. The detection was solved using the You Look Only Once (YOLO) model, which works on the principle of a convolutional neural network. The OCR tool PythonOCR was used to read characters from tags. The success rate of IS21 tag detection is 93.4%, and the success rate of reading text from tags is equal to 85.9%. The architecture described in the article is suitable for solving the defined problem. Springer Science and Business Media Deutschland GmbH 2022 Article PeerReviewed Kopecky, Lukas and Dobrovolny, Michal and Fuchs, Antonin and Selamat, Ali and Krejcar, Ondrej (2022) Cycle route signs detection using deep learning. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 13501 (NA). pp. 82-94. ISSN 0302-9743 http://dx.doi.org/10.1007/978-3-031-16014-1_8 DOI : 10.1007/978-3-031-16014-1_8 |
institution |
Universiti Teknologi Malaysia |
building |
UTM Library |
collection |
Institutional Repository |
continent |
Asia |
country |
Malaysia |
content_provider |
Universiti Teknologi Malaysia |
content_source |
UTM Institutional Repository |
url_provider |
http://eprints.utm.my/ |
topic |
QA76 Computer software |
spellingShingle |
QA76 Computer software Kopecky, Lukas Dobrovolny, Michal Fuchs, Antonin Selamat, Ali Krejcar, Ondrej Cycle route signs detection using deep learning |
description |
This article addresses the issue of detecting traffic signs signalling cycle routes. It is also necessary to read the number or text of the cycle route from the given image. These tags are kept under the identifier IS21 and have a defined, uniform design with text in the middle of the tag. The detection was solved using the You Look Only Once (YOLO) model, which works on the principle of a convolutional neural network. The OCR tool PythonOCR was used to read characters from tags. The success rate of IS21 tag detection is 93.4%, and the success rate of reading text from tags is equal to 85.9%. The architecture described in the article is suitable for solving the defined problem. |
format |
Article |
author |
Kopecky, Lukas Dobrovolny, Michal Fuchs, Antonin Selamat, Ali Krejcar, Ondrej |
author_facet |
Kopecky, Lukas Dobrovolny, Michal Fuchs, Antonin Selamat, Ali Krejcar, Ondrej |
author_sort |
Kopecky, Lukas |
title |
Cycle route signs detection using deep learning |
title_short |
Cycle route signs detection using deep learning |
title_full |
Cycle route signs detection using deep learning |
title_fullStr |
Cycle route signs detection using deep learning |
title_full_unstemmed |
Cycle route signs detection using deep learning |
title_sort |
cycle route signs detection using deep learning |
publisher |
Springer Science and Business Media Deutschland GmbH |
publishDate |
2022 |
url |
http://eprints.utm.my/id/eprint/100514/ http://dx.doi.org/10.1007/978-3-031-16014-1_8 |
_version_ |
1764222577724620800 |
score |
13.154949 |