A Text Recognition Dataset from Sahidic Coptic Ancient Manuscripts

2026-06-14

AI Summary

This study addresses the challenges of handwritten text recognition (HTR) in low-resource settings, particularly concerning endangered languages, rare scripts, and degraded historical documents. To this end, we introduce SCAM, the first line-level image dataset dedicated to Sahidic Coptic manuscripts, encompassing materials from multiple collections and exhibiting typical forms of visual degradation. Sahidic Coptic, an extinct language with a unique alphabet and distinctive diacritical marks, represents an extremely scarce linguistic resource. We conduct comprehensive benchmarking of state-of-the-art HTR models on SCAM, systematically evaluating their performance under conditions of limited training data and high visual complexity. Our experiments reveal a significant performance gap between current methods trained on modern, resource-rich scripts and those applied to historical, low-resource scenarios, thereby establishing a crucial foundation for future research through both data and evaluation benchmarks.

Abstract

In this work, we target Handwritten Text Recognition (HTR) in low-resource scenarios, which arise from underrepresented languages, rare scripts, and degraded visual conditions typical of historical documents. We introduce SCAM (Sahidic Coptic Ancient Manuscripts), a new line-level dataset built from digitized ancient manuscripts written in the extinct Sahidic Coptic dialect. The dataset reflects a realistic and challenging setting, as it combines heterogeneous acquisition conditions across libraries with typical manuscript degradations such as ink fading, bleed-through, and material deterioration. In addition to visual complexity, SCAM poses significant linguistic challenges due to the scarcity of resources for Sahidic Coptic, its uncommon alphabet, and dialect-specific diacritics. To support research in low-resource HTR, we benchmark several state-of-the-art approaches based on different paradigms, highlighting their limitations and strengths in this setting. Our results underline the gap between current HTR performance on well-resourced modern scripts and historically grounded, low-resource scenarios, thus providing a reference point for future developments.

Problem

• Handwritten Text Recognition
• low-resource
• historical manuscripts
• Sahidic Coptic
• script recognition

Innovation

• Handwritten Text Recognition
• low-resource
• Sahidic Coptic
• historical manuscripts
• dataset