ICDAR 2026 HIPE-OCRepair Competition on LLM-Assisted OCR Post-Correction for Historical Documents

2026-07-09

AI Summary

This work addresses the challenge of textual errors introduced by legacy OCR systems in the digitization of historical documents, proposing an efficient post-correction approach that obviates the need for rescanning. By constructing HIPE-OCRepair-2026—a multilingual historical document dataset—the study presents the first systematic evaluation of large language models (LLMs) for paragraph-level OCR post-correction on 17th–20th-century English, French, and German printed texts without access to original images. It examines diverse strategies including zero-shot prompting, continued pretraining, and fine-tuning, and introduces a novel retrieval-oriented evaluation paradigm alongside a unified, reproducible benchmark framework. Experimental results demonstrate that LLMs substantially improve OCR quality, yet their performance is modulated by factors such as language and noise level, with a tendency toward over-correction under low-noise conditions—highlighting the necessity of holistic evaluation metrics beyond character error rate.

Abstract

We present the results of HIPE-OCRepair-2026, an ICDAR competition on LLM-assisted OCR post-correction of historical documents. OCR post-correction remains a long-standing challenge in digital heritage: large-scale collections of digitized documents are affected by legacy OCR errors, while re-digitization at scale remains impractical. Large language models (LLMs) offers a major opportunity to revisit this challenge, yet their effectiveness across languages, document types, and noise conditions - and their tendency to hallucinate - remains insufficiently understood. HIPE-OCRepair-2026 pursues two objectives: (i) to evaluate the capabilities of modern OCR post-correction systems, and (ii) to provide a reproducible evaluation framework anchored in the HIPE-OCRepair-2026 dataset, a harmonized multilingual resource consolidating existing and newly curated historical datasets. Participants were tasked with correcting noisy OCR transcripts from historical newspapers and printed works in English, French, and German (17th-20th century), working at the level of coherent transcription units (paragraphs or articles) without access to source images. The evaluation adopts a retrieval-oriented rather than diplomatic scoring approach, reflecting the practical use case of search and access over digitized collections. Four teams submitted systems ranging from zero-shot prompting to continued pre-training and fine-tuning, offering insights into the merits of different adaptation strategies. Results show that modern LLM-assisted systems can significantly improve OCR quality, but performance varies across datasets, languages, and noise levels. Over-correction on low-noise inputs emerges as a recurring challenge, highlighting the importance of evaluation beyond character error reduction. The dataset, scorer, and evaluation pipeline are publicly released to support future research.

Problem

• OCR post-correction
• historical documents
• digital heritage
• LLM hallucination
• multilingual OCR

Innovation

• LLM-assisted OCR
• historical document processing
• multilingual post-correction
• retrieval-oriented evaluation
• HIPE-OCRepair