Can one deep learning model learn script-independent multilingual word-spotting?

Mohammed Al-Rawi; Ernest Valveny; Dimosthenis Karatzas

doi:10.1109/ICDAR.2019.00050

Can one deep learning model learn script-independent multilingual word-spotting?

Mohammed Al-Rawi, Ernest Valveny, Dimosthenis Karatzas

Department of Computer Science

Computer Vision Center (CVC)

Research output: Chapter in Book › Chapter › Research › peer-review

3 Citations (Scopus)

Abstract

Word spotting has gained increased attention lately as it can be used to extract textual information from handwritten documents and scene-text images. Current word spotting approaches are designed to work on a single language and/or script. Building intelligent models that learn script-independent multilingual word-spotting is challenging due to the large variability of multilingual alphabets and symbols. We used ResNet-152 and the Pyramidal Histogram of Characters (PHOC) embedding to build a one-model script-independent multilingual word-spotting and we tested it on Latin, Arabic, and Bangla (Indian) languages. The one-model we propose performs on par with the multi-model language-specific word-spotting system, and thus, reduces the number of models needed for each script and/or language.

Original language	English
Title of host publication	Proceedings - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019
Pages	260-267
Number of pages	8
ISBN (Electronic)	9781728128610
DOIs	https://doi.org/10.1109/ICDAR.2019.00050
Publication status	Published - Sept 2019

Publication series

Name	Proceedings of the International Conference on Document Analysis and Recognition, ICDAR
ISSN (Print)	1520-5363

Keywords

Handwriting
Histogram of Characters
Multitasking
PHOC
ResNet
Scene text images

Access to Document

10.1109/ICDAR.2019.00050

Cite this

Al-Rawi, M., Valveny, E., & Karatzas, D. (2019). Can one deep learning model learn script-independent multilingual word-spotting? In Proceedings - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019 (pp. 260-267). Article 8978043 (Proceedings of the International Conference on Document Analysis and Recognition, ICDAR). https://doi.org/10.1109/ICDAR.2019.00050

@inbook{a4e2820fa88744b0948021c6c0a7a9c6,

title = "Can one deep learning model learn script-independent multilingual word-spotting?",

abstract = "Word spotting has gained increased attention lately as it can be used to extract textual information from handwritten documents and scene-text images. Current word spotting approaches are designed to work on a single language and/or script. Building intelligent models that learn script-independent multilingual word-spotting is challenging due to the large variability of multilingual alphabets and symbols. We used ResNet-152 and the Pyramidal Histogram of Characters (PHOC) embedding to build a one-model script-independent multilingual word-spotting and we tested it on Latin, Arabic, and Bangla (Indian) languages. The one-model we propose performs on par with the multi-model language-specific word-spotting system, and thus, reduces the number of models needed for each script and/or language.",

keywords = "Handwriting, Histogram of Characters, Multitasking, PHOC, ResNet, Scene text images",

author = "Mohammed Al-Rawi and Ernest Valveny and Dimosthenis Karatzas",

note = "Funding Information: This work has received funding from from the European Union's Horizon 2020 research and innovation programme under the Marie Skodowska-Curie grant agreement No 665919, research grants TIN2017-89779-P and 2017-SGR-1783 from the Spanish and Catalan government, respectively. Publisher Copyright: {\textcopyright} 2019 IEEE.",

year = "2019",

month = sep,

doi = "10.1109/ICDAR.2019.00050",

language = "English",

series = "Proceedings of the International Conference on Document Analysis and Recognition, ICDAR",

pages = "260--267",

booktitle = "Proceedings - 15th IAPR International Conference on Document Analysis and Recognition, ICDAR 2019",

}