Localization of visual codes in the DCT domain using deep rectifier neural networks

Péter Bodnár, Tamás Grósz, László Tóth, L. Nyúl

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

The reading process of visual codes consists of two steps, localization and data decoding. This paper presents a novel method for QR code localization using deep rectifier neural networks, trained directly in the JPEG DCT domain, thus making image decompression unnecessary. This approach is efficient with respect to both storage and computation cost, being convenient, since camera hardware can provide JPEG stream as their output in many cases. The structure of the neural networks, regularization, and training data parameters, like input vector length and compression level, are evaluated and discussed. The proposed approach is not exclusively for QR codes, but can be adapted to Data Matrix codes or other two-dimensional code types as well.

Original languageEnglish
Title of host publicationProceedings of the International Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjunction with ICINCO 2014
PublisherSciTePress
Pages37-44
Number of pages8
ISBN (Print)9789897580413
Publication statusPublished - 2014
EventInternational Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjuction with the International Conference on Informatics in Control, Automation and Robotics, ICINCO 2014 - Vienna, Austria
Duration: Sep 1 2014Sep 3 2014

Other

OtherInternational Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjuction with the International Conference on Informatics in Control, Automation and Robotics, ICINCO 2014
CountryAustria
CityVienna
Period9/1/149/3/14

Fingerprint

Neural networks
Decoding
Cameras
Hardware
Costs

ASJC Scopus subject areas

  • Artificial Intelligence
  • Information Systems

Cite this

Bodnár, P., Grósz, T., Tóth, L., & Nyúl, L. (2014). Localization of visual codes in the DCT domain using deep rectifier neural networks. In Proceedings of the International Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjunction with ICINCO 2014 (pp. 37-44). SciTePress.

Localization of visual codes in the DCT domain using deep rectifier neural networks. / Bodnár, Péter; Grósz, Tamás; Tóth, László; Nyúl, L.

Proceedings of the International Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjunction with ICINCO 2014. SciTePress, 2014. p. 37-44.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Bodnár, P, Grósz, T, Tóth, L & Nyúl, L 2014, Localization of visual codes in the DCT domain using deep rectifier neural networks. in Proceedings of the International Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjunction with ICINCO 2014. SciTePress, pp. 37-44, International Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjuction with the International Conference on Informatics in Control, Automation and Robotics, ICINCO 2014, Vienna, Austria, 9/1/14.
Bodnár P, Grósz T, Tóth L, Nyúl L. Localization of visual codes in the DCT domain using deep rectifier neural networks. In Proceedings of the International Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjunction with ICINCO 2014. SciTePress. 2014. p. 37-44
Bodnár, Péter ; Grósz, Tamás ; Tóth, László ; Nyúl, L. / Localization of visual codes in the DCT domain using deep rectifier neural networks. Proceedings of the International Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjunction with ICINCO 2014. SciTePress, 2014. pp. 37-44
@inproceedings{40ab10cb9cea468b959b3bd011c80b99,
title = "Localization of visual codes in the DCT domain using deep rectifier neural networks",
abstract = "The reading process of visual codes consists of two steps, localization and data decoding. This paper presents a novel method for QR code localization using deep rectifier neural networks, trained directly in the JPEG DCT domain, thus making image decompression unnecessary. This approach is efficient with respect to both storage and computation cost, being convenient, since camera hardware can provide JPEG stream as their output in many cases. The structure of the neural networks, regularization, and training data parameters, like input vector length and compression level, are evaluated and discussed. The proposed approach is not exclusively for QR codes, but can be adapted to Data Matrix codes or other two-dimensional code types as well.",
author = "P{\'e}ter Bodn{\'a}r and Tam{\'a}s Gr{\'o}sz and L{\'a}szl{\'o} T{\'o}th and L. Ny{\'u}l",
year = "2014",
language = "English",
isbn = "9789897580413",
pages = "37--44",
booktitle = "Proceedings of the International Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjunction with ICINCO 2014",
publisher = "SciTePress",

}

TY - GEN

T1 - Localization of visual codes in the DCT domain using deep rectifier neural networks

AU - Bodnár, Péter

AU - Grósz, Tamás

AU - Tóth, László

AU - Nyúl, L.

PY - 2014

Y1 - 2014

N2 - The reading process of visual codes consists of two steps, localization and data decoding. This paper presents a novel method for QR code localization using deep rectifier neural networks, trained directly in the JPEG DCT domain, thus making image decompression unnecessary. This approach is efficient with respect to both storage and computation cost, being convenient, since camera hardware can provide JPEG stream as their output in many cases. The structure of the neural networks, regularization, and training data parameters, like input vector length and compression level, are evaluated and discussed. The proposed approach is not exclusively for QR codes, but can be adapted to Data Matrix codes or other two-dimensional code types as well.

AB - The reading process of visual codes consists of two steps, localization and data decoding. This paper presents a novel method for QR code localization using deep rectifier neural networks, trained directly in the JPEG DCT domain, thus making image decompression unnecessary. This approach is efficient with respect to both storage and computation cost, being convenient, since camera hardware can provide JPEG stream as their output in many cases. The structure of the neural networks, regularization, and training data parameters, like input vector length and compression level, are evaluated and discussed. The proposed approach is not exclusively for QR codes, but can be adapted to Data Matrix codes or other two-dimensional code types as well.

UR - http://www.scopus.com/inward/record.url?scp=84908868494&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84908868494&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84908868494

SN - 9789897580413

SP - 37

EP - 44

BT - Proceedings of the International Workshop on Artificial Neural Networks and Intelligent Information Processing, ANNIIP 2014 - In Conjunction with ICINCO 2014

PB - SciTePress

ER -