Importance of Textlines in Historical Document Classification

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F22%3APU144693" target="_blank" >RIV/00216305:26230/22:PU144693 - isvavai.cz</a>

Výsledek na webu

<a href="https://pero.fit.vutbr.cz/publications" target="_blank" >https://pero.fit.vutbr.cz/publications</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/978-3-031-06555-2_11" target="_blank" >10.1007/978-3-031-06555-2_11</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Importance of Textlines in Historical Document Classification

Popis výsledku v původním jazyce

This paper describes a system prepared at Brno University of Technology for ICDAR 2021 Competition on Historical Document Classification, experiments leading to its design, and the main findings. The solved tasks include script and font classification, document origin localization, and dating. We combined patch-level and line-level approaches, where the line-level system utilizes an existing, publicly available page layout analysis engine. In both systems, neural networks provide local predictions which are combined into page-level decisions, and the results of both systems are fused using linear or log-linear interpolation. We propose loss functions suitable for weakly supervised classification problem where multiple possible labels are provided, and we propose loss functions suitable for interval regression in the dating task. The line-level system significantly improves results in script and font classification and in the dating task. The full system achieved 98.48%, 88.84%, and 79.69% accuracy in the font, script, and location classification tasks respectively. In the dating task, our system achieved a mean absolute error of 21.91 years. Our system achieved the best results in all tasks and became the overall winner of the competition.

Název v anglickém jazyce

Importance of Textlines in Historical Document Classification

Popis výsledku anglicky

This paper describes a system prepared at Brno University of Technology for ICDAR 2021 Competition on Historical Document Classification, experiments leading to its design, and the main findings. The solved tasks include script and font classification, document origin localization, and dating. We combined patch-level and line-level approaches, where the line-level system utilizes an existing, publicly available page layout analysis engine. In both systems, neural networks provide local predictions which are combined into page-level decisions, and the results of both systems are fused using linear or log-linear interpolation. We propose loss functions suitable for weakly supervised classification problem where multiple possible labels are provided, and we propose loss functions suitable for interval regression in the dating task. The line-level system significantly improves results in script and font classification and in the dating task. The full system achieved 98.48%, 88.84%, and 79.69% accuracy in the font, script, and location classification tasks respectively. In the dating task, our system achieved a mean absolute error of 21.91 years. Our system achieved the best results in all tasks and became the overall winner of the competition.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

Výsledek vznikl pri realizaci vícero projektů. Více informací v záložce Projekty.

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2022

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Uchida, S., Barney, E., Eglin, V. (eds) Document Analysis Systems

ISBN

978-3-031-06554-5

ISSN

—

e-ISSN

—

Počet stran výsledku

13

Strana od-do

158-170

Název nakladatele

Springer Nature Switzerland AG

Místo vydání

La Rochelle

Místo konání akce

La Rochelle

Datum konání akce

22. 5. 2022

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

000870314500011