Improving Automatic Text Recognition with Language Models in the PyLaia Open-Source Library

TL;DR

This paper enhances the PyLaia OCR library by integrating auto-tuned language models and confidence scoring, significantly improving recognition accuracy across multiple datasets with minimal expert intervention.

Contribution

It introduces an easy-to-use, auto-tuned language modeling integration in PyLaia, boosting OCR performance without requiring additional data or expert knowledge.

Findings

Word Error Rate reduced by 13% on average

Character Error Rate reduced by 12% on average

Confidence score calibration improves reliability

Abstract

PyLaia is one of the most popular open-source software for Automatic Text Recognition (ATR), delivering strong performance in terms of speed and accuracy. In this paper, we outline our recent contributions to the PyLaia library, focusing on the incorporation of reliable confidence scores and the integration of statistical language modeling during decoding. Our implementation provides an easy way to combine PyLaia with n-grams language models at different levels. One of the highlights of this work is that language models are completely auto-tuned: they can be built and used easily without any expert knowledge, and without requiring any additional data. To demonstrate the significance of our contribution, we evaluate PyLaia's performance on twelve datasets, both with and without language modelling. The results show that decoding with small language models improves the Word Error Rate by 13% and the Character Error Rate by 12% in average. Additionally, we conduct an analysis of confidence scores and highlight the importance of calibration techniques. Our implementation is publicly available in the official PyLaia repository at https://gitlab.teklia.com/atr/pylaia, and twelve open-source models are released on Hugging Face.