Cross-Temporal Sinhala OCR: Page-Level Adaptation and Diachronic Analysis

2026-06-28

AI Summary

Existing Sinhala full-page OCR research is hindered by the scarcity of real-world datasets, often relying on synthetic data that fails to capture the diversity and degradation present in actual printed documents. To address this gap, this work introduces Sinhala-OCR-LK-Acts-1010, the first real-world Sinhala OCR dataset, and employs QLoRA for efficient fine-tuning of vision-language models such as DeepSeek-OCR V1/V2 and LightOnOCR-2-1B. Experimental results demonstrate that LightOnOCR-2-1B achieves robust performance on cross-era degraded documents, attaining a character error rate as low as 1.05%, significantly outperforming Surya-OCR (8.84%), Tesseract v5 (10.69%), and Google Document AI (2.06%). These findings underscore the effectiveness of lightweight large models in low-resource OCR scenarios.

Abstract

Sinhala is a morphologically rich abugida spoken by roughly 16 million people in Sri Lanka, and to date, there are no publicly available real-world datasets for page-level Sinhala OCR. All previous studies for assessing Sinhala OCR models have used artificially generated data. To bridge the gap, we introduce sinhala-ocr-lk-acts-1010, an annotated dataset of 1,010 page-level images and their transcriptions collected from Sri Lankan Legislative Acts published between 1981-1989 and 2000-2019, split into 707 training examples, 101 validation examples, and 202 testing examples. Three models based on deep learning-based visual language processing, namely DeepSeek-OCR V1, DeepSeek-OCR V2, and LightOnOCR-2-1B, are fine-tuned using QLoRA in 8 experiments conducted on consumer and cloud GPUs. LightOnOCR-2-1B is the top performer, achieving a CER of 1.05% across all test examples, outperforming state-of-the-art open-source OCR models such as Surya-OCR (8.84%) and Tesseract v5 (10.69%), as well as commercially available OCR models such as Google Document AI (2.06%). Our results suggest that LightOnOCR-2-1B outperforms other baselines on real-world OCR tasks and maintains consistent performance across all print periods, even when documents are severely degraded.

Problem

• Sinhala OCR
• page-level dataset
• real-world documents
• diachronic analysis
• optical character recognition

Innovation

• Sinhala OCR
• page-level dataset
• QLoRA fine-tuning
• diachronic analysis
• LightOnOCR-2-1B