Mixed Model OCR Training on Historical Latin Script for Out-of-the-Box Recognition and Finetuning

TL;DR

This paper presents a robust polyfont OCR model for historical Latin scripts that achieves low error rates out-of-the-box and can be efficiently fine-tuned for specific printings, enhancing automatic recognition of diverse historical texts.

Contribution

The authors develop a widely applicable mixed-model OCR for Latin scripts that outperforms standard models and simplifies fine-tuning for specific historical printings.

Findings

Achieved CER of 1.73% on unseen books, outperforming standard models.

Fine-tuning from the mixed model reduces CER by up to 50%.

The model is openly available for community use.

Abstract

In order to apply Optical Character Recognition (OCR) to historical printings of Latin script fully automatically, we report on our efforts to construct a widely-applicable polyfont recognition model yielding text with a Character Error Rate (CER) around 2% when applied out-of-the-box. Moreover, we show how this model can be further finetuned to specific classes of printings with little manual and computational effort. The mixed or polyfont model is trained on a wide variety of materials, in terms of age (from the 15th to the 19th century), typography (various types of Fraktur and Antiqua), and languages (among others, German, Latin, and French). To optimize the results we combined established techniques of OCR training like pretraining, data augmentation, and voting. In addition, we used various preprocessing methods to enrich the training data and obtain more robust models. We also implemented a two-stage approach which first trains on all available, considerably unbalanced data and then refines the output by training on a selected more balanced subset. Evaluations on 29 previously unseen books resulted in a CER of 1.73%, outperforming a widely used standard model with a CER of 2.84% by almost 40%. Training a more specialized model for some unseen Early Modern Latin books starting from our mixed model led to a CER of 1.47%, an improvement of up to 50% compared to training from scratch and up to 30% compared to training from the aforementioned standard model. Our new mixed model is made openly available to the community.