Synthesizing versus Augmentation for Arabic Word Recognition with Convolutional Neural Networks

Reem Alaasam; Berat Kurar Barakat; Jihad El-Sana

doi:10.1109/ASAR.2018.8480189

Synthesizing versus Augmentation for Arabic Word Recognition with Convolutional Neural Networks

Reem Alaasam, Berat Kurar Barakat, Jihad El-Sana

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

In this paper, we present a sub-word recognition method for historical Arabic manuscripts, using convolutional neural networks. We investigate the benefit of extending training set with synthetically created samples in comparison to augmentation. We show that annotating around ten pages of a manuscript and extending it, is sufficient for successful sub-word recognition in the whole manuscript. In addition, we show the contribution of using different combinations of training sets and compare their sub-word recognition performance in the whole manuscript.

Original language	English
Title of host publication	2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	114-118
Number of pages	5
ISBN (Electronic)	9781538614594
DOIs	https://doi.org/10.1109/ASAR.2018.8480189
State	Published - 2 Oct 2018
Externally published	Yes
Event	2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018 - London, United Kingdom Duration: 12 Mar 2018 → 14 Mar 2018

Publication series

Name	2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018

Conference

Conference	2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018
Country/Territory	United Kingdom
City	London
Period	12/03/18 → 14/03/18

Keywords

Arabic
Database
handwritten
text recognition

Access to Document

10.1109/ASAR.2018.8480189

Cite this

Alaasam, R., Barakat, B. K., & El-Sana, J. (2018). Synthesizing versus Augmentation for Arabic Word Recognition with Convolutional Neural Networks. In 2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018 (pp. 114-118). Article 8480189 (2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ASAR.2018.8480189

Alaasam, Reem ; Barakat, Berat Kurar ; El-Sana, Jihad. / Synthesizing versus Augmentation for Arabic Word Recognition with Convolutional Neural Networks. 2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018. Institute of Electrical and Electronics Engineers Inc., 2018. pp. 114-118 (2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018).

@inproceedings{98fa521fb90347739d8bf66511cd78d7,

title = "Synthesizing versus Augmentation for Arabic Word Recognition with Convolutional Neural Networks",

abstract = "In this paper, we present a sub-word recognition method for historical Arabic manuscripts, using convolutional neural networks. We investigate the benefit of extending training set with synthetically created samples in comparison to augmentation. We show that annotating around ten pages of a manuscript and extending it, is sufficient for successful sub-word recognition in the whole manuscript. In addition, we show the contribution of using different combinations of training sets and compare their sub-word recognition performance in the whole manuscript.",

keywords = "Arabic, Database, handwritten, text recognition",

author = "Reem Alaasam and Barakat, {Berat Kurar} and Jihad El-Sana",

note = "Publisher Copyright: {\textcopyright} 2018 IEEE.; 2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018 ; Conference date: 12-03-2018 Through 14-03-2018",

year = "2018",

month = oct,

day = "2",

doi = "10.1109/ASAR.2018.8480189",

language = "אנגלית",

series = "2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "114--118",

booktitle = "2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018",

address = "ארצות הברית",

}

Alaasam, R, Barakat, BK & El-Sana, J 2018, Synthesizing versus Augmentation for Arabic Word Recognition with Convolutional Neural Networks. in 2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018., 8480189, 2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018, Institute of Electrical and Electronics Engineers Inc., pp. 114-118, 2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018, London, United Kingdom, 12/03/18. https://doi.org/10.1109/ASAR.2018.8480189

Synthesizing versus Augmentation for Arabic Word Recognition with Convolutional Neural Networks. / Alaasam, Reem; Barakat, Berat Kurar; El-Sana, Jihad.
2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018. Institute of Electrical and Electronics Engineers Inc., 2018. p. 114-118 8480189 (2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Synthesizing versus Augmentation for Arabic Word Recognition with Convolutional Neural Networks

AU - Alaasam, Reem

AU - Barakat, Berat Kurar

AU - El-Sana, Jihad

PY - 2018/10/2

Y1 - 2018/10/2

N2 - In this paper, we present a sub-word recognition method for historical Arabic manuscripts, using convolutional neural networks. We investigate the benefit of extending training set with synthetically created samples in comparison to augmentation. We show that annotating around ten pages of a manuscript and extending it, is sufficient for successful sub-word recognition in the whole manuscript. In addition, we show the contribution of using different combinations of training sets and compare their sub-word recognition performance in the whole manuscript.

AB - In this paper, we present a sub-word recognition method for historical Arabic manuscripts, using convolutional neural networks. We investigate the benefit of extending training set with synthetically created samples in comparison to augmentation. We show that annotating around ten pages of a manuscript and extending it, is sufficient for successful sub-word recognition in the whole manuscript. In addition, we show the contribution of using different combinations of training sets and compare their sub-word recognition performance in the whole manuscript.

KW - Arabic

KW - Database

KW - handwritten

KW - text recognition

UR - http://www.scopus.com/inward/record.url?scp=85056184786&partnerID=8YFLogxK

U2 - 10.1109/ASAR.2018.8480189

DO - 10.1109/ASAR.2018.8480189

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:85056184786

T3 - 2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018

SP - 114

EP - 118

BT - 2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018

Y2 - 12 March 2018 through 14 March 2018

ER -

Alaasam R, Barakat BK, El-Sana J. Synthesizing versus Augmentation for Arabic Word Recognition with Convolutional Neural Networks. In 2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018. Institute of Electrical and Electronics Engineers Inc. 2018. p. 114-118. 8480189. (2nd IEEE International Workshop on Arabic and Derived Script Analysis and Recognition, ASAR 2018). doi: 10.1109/ASAR.2018.8480189

Synthesizing versus Augmentation for Arabic Word Recognition with Convolutional Neural Networks

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this