Evaluating State‐of‐the‐Art Handwritten Text Recognition (HTR) Engines; with Large Language Models (LLMs) for Historical Document Digitisation

Gundram Leifert; Christel Annemieke Romein; Achim Rabus; Phillip Benjamin Ströbel; Benjamin Kiessling; Tobias Hödel

doi:10.5281/ZENODO.8102666

Evaluating State‐of‐the‐Art Handwritten Text Recognition (HTR) Engines; with Large Language Models (LLMs) for Historical Document Digitisation

Gundram Leifert, Christel Annemieke Romein^*, Achim Rabus, Phillip Benjamin Ströbel, Benjamin Kiessling, Tobias Hödel

^*Corresponding author for this work

Huygens - Politieke Cultuur en Geschiedenis

Research output: Contribution to conference › Poster › Scientific

Abstract

Handwritten Text Recognition engines play a pivotal role in transforming historical handwritten documents into machine‐readable formats, facilitating efficient access and analysis of invaluable information. We present a comprehensive evaluation of cutting‐edge Handwritten Text Recognition engines, including PyLaia, IDA, and TrOCR. The study aims to assist researchers in making informed decisions for datafication of historical documents. We ensure a fair comparison of the engine performance by utilizing diverse datasets comprising varying languages, scripts, complexities, and historical periods. As a benchmark for improvement, we incorporate datasets previously processed with Transkribus/CitLab HTR+, allowing us to assess the engines' recognition accuracy. By analysing the Character Error Rate (CER), we evaluate their performance.

Original language	English
DOIs	https://doi.org/10.5281/ZENODO.8102666
Publication status	Published - 07 Dec 2023

Keywords

LLMs
HTR

Access to Document

10.5281/ZENODO.8102666

Embargoed Document

Poster_HTR_CHR_Paris2023
Final published version, 2.01 MB
Request copy

Cite this

@conference{f16b1edff5c8443e8b696c272b9aa72f,

title = "Evaluating State‐of‐the‐Art Handwritten Text Recognition (HTR) Engines; with Large Language Models (LLMs) for Historical Document Digitisation",

abstract = "Handwritten Text Recognition engines play a pivotal role in transforming historical handwritten documents into machine‐readable formats, facilitating efficient access and analysis of invaluable information. We present a comprehensive evaluation of cutting‐edge Handwritten Text Recognition engines, including PyLaia, IDA, and TrOCR. The study aims to assist researchers in making informed decisions for datafication of historical documents. We ensure a fair comparison of the engine performance by utilizing diverse datasets comprising varying languages, scripts, complexities, and historical periods. As a benchmark for improvement, we incorporate datasets previously processed with Transkribus/CitLab HTR+, allowing us to assess the engines' recognition accuracy. By analysing the Character Error Rate (CER), we evaluate their performance.",

keywords = "LLMs, HTR",

author = "Gundram Leifert and Romein, {Christel Annemieke} and Achim Rabus and Str{\"o}bel, {Phillip Benjamin} and Benjamin Kiessling and Tobias H{\"o}del",

year = "2023",

month = dec,

day = "7",

doi = "10.5281/ZENODO.8102666",

language = "English",

}

TY - CONF

T1 - Evaluating State‐of‐the‐Art Handwritten Text Recognition (HTR) Engines; with Large Language Models (LLMs) for Historical Document Digitisation

AU - Leifert, Gundram

AU - Romein, Christel Annemieke

AU - Rabus, Achim

AU - Ströbel, Phillip Benjamin

AU - Kiessling, Benjamin

AU - Hödel, Tobias

PY - 2023/12/7

Y1 - 2023/12/7

N2 - Handwritten Text Recognition engines play a pivotal role in transforming historical handwritten documents into machine‐readable formats, facilitating efficient access and analysis of invaluable information. We present a comprehensive evaluation of cutting‐edge Handwritten Text Recognition engines, including PyLaia, IDA, and TrOCR. The study aims to assist researchers in making informed decisions for datafication of historical documents. We ensure a fair comparison of the engine performance by utilizing diverse datasets comprising varying languages, scripts, complexities, and historical periods. As a benchmark for improvement, we incorporate datasets previously processed with Transkribus/CitLab HTR+, allowing us to assess the engines' recognition accuracy. By analysing the Character Error Rate (CER), we evaluate their performance.

AB - Handwritten Text Recognition engines play a pivotal role in transforming historical handwritten documents into machine‐readable formats, facilitating efficient access and analysis of invaluable information. We present a comprehensive evaluation of cutting‐edge Handwritten Text Recognition engines, including PyLaia, IDA, and TrOCR. The study aims to assist researchers in making informed decisions for datafication of historical documents. We ensure a fair comparison of the engine performance by utilizing diverse datasets comprising varying languages, scripts, complexities, and historical periods. As a benchmark for improvement, we incorporate datasets previously processed with Transkribus/CitLab HTR+, allowing us to assess the engines' recognition accuracy. By analysing the Character Error Rate (CER), we evaluate their performance.

KW - LLMs

KW - HTR

U2 - 10.5281/ZENODO.8102666

DO - 10.5281/ZENODO.8102666

M3 - Poster

ER -

Evaluating State‐of‐the‐Art Handwritten Text Recognition (HTR) Engines; with Large Language Models (LLMs) for Historical Document Digitisation

Abstract

Keywords

Access to Document

Embargoed Document

Fingerprint

Cite this