# Ground Truth for training OCR engines on historical documents in German Fraktur and Early Modern Latin

arXiv: https://arxiv.org/abs/1809.05501 · HTML: https://mlsanity.com/paper/1809.05501/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 14 Sep 2018.

## Built on by (most cited first, 12 of 12)

- [Optical Character Recognition of 19th Century Classical Commentaries: the Current State of Affairs](https://mlsanity.com/paper/2110.06817/) (2021)
- [Mixed Model OCR Training on Historical Latin Script for Out-of-the-Box Recognition and Finetuning](https://mlsanity.com/paper/2106.07881/) (2021)
- [Detection Masking for Improved OCR on Noisy Documents](https://mlsanity.com/paper/2205.08257/) (2022)
- [Evaluating Robustness of LLMs in Question Answering on Multilingual Noisy OCR Data](https://mlsanity.com/paper/2502.16781/) (2025)
- [Error Patterns in Historical OCR: A Comparative Analysis of TrOCR and a Vision-Language Model](https://mlsanity.com/paper/2602.14524/) (2026)
- [A Survey of OCR Evaluation Methods and Metrics and the Invisibility of Historical Documents](https://mlsanity.com/paper/2603.25761/) (2026)
- [BusiNet -- a Light and Fast Text Detection Network for Business Documents](https://mlsanity.com/paper/2207.01220/) (2022)
- [Combining OCR Models for Reading Early Modern Printed Books](https://mlsanity.com/paper/2305.07131/) (2023)
- [Advances and Limitations in Open Source Arabic-Script OCR: A Case Study](https://mlsanity.com/paper/2402.10943/) (2024)
- [Making Old Kurdish Publications Processable by Augmenting Available Optical Character Recognition Engines](https://mlsanity.com/paper/2404.06101/) (2024)
- [Chronos: The AI Co-Historian](https://mlsanity.com/paper/2604.03553/) (2026)
- [ICDAR 2026 HIPE-OCRepair Competition on LLM-Assisted OCR Post-Correction for Historical Documents](https://mlsanity.com/paper/2607.08143/) (2026)

## Abstract (arXiv)

In this paper we describe a dataset of German and Latin textitground truth (GT) for historical OCR in the form of printed text line images paired with their transcription. This dataset, called textitGT4HistOCR, consists of 313,173 line pairs covering a wide period of printing dates from incunabula from the 15th century to 19th century books printed in Fraktur types and is openly available under a CC-BY 4.0 license. The special form of GT as line image/transcription pairs makes it directly usable to train state-of-the-art recognition models for OCR software employing recurring neural networks in LSTM architecture such as Tesseract 4 or OCRopus. We also provide some pretrained OCRopus models for subcorpora of our dataset yielding between 95% (early printings) and 98% (19th century Fraktur printings) character accuracy rates on unseen test cases, a Perl script to harmonize GT produced by different transcription rules, and give hints on how to construct GT for OCR purposes which has requirements that may differ from linguistically motivated transcriptions.