OCR Improves Machine Translation for Low-Resource Languages

TL;DR

This paper evaluates OCR systems on low-resource languages, introduces a new benchmark, and demonstrates how OCR data can enhance machine translation performance through backtranslation, highlighting the importance of OCR quality.

Contribution

It presents OCR4MT, a novel benchmark for low-resource languages, and analyzes OCR errors' impact on machine translation, proposing optimal OCR quality levels for effective use.

Findings

OCR monolingual data improves MT performance via backtranslation

OCR errors negatively affect MT quality

A minimum OCR quality threshold is identified for usefulness

Abstract

We aim to investigate the performance of current OCR systems on low resource languages and low resource scripts. We introduce and make publicly available a novel benchmark, OCR4MT, consisting of real and synthetic data, enriched with noise, for 60 low-resource languages in low resource scripts. We evaluate state-of-the-art OCR systems on our benchmark and analyse most common errors. We show that OCR monolingual data is a valuable resource that can increase performance of Machine Translation models, when used in backtranslation. We then perform an ablation study to investigate how OCR errors impact Machine Translation performance and determine what is the minimum level of OCR quality needed for the monolingual data to be useful for Machine Translation.