On the Accuracy of CRNNs for Line-Based OCR: A Multi-Parameter Evaluation

TL;DR

This paper evaluates how various parameters affect CRNN-based OCR accuracy on difficult historical texts, optimizing architecture and training settings to achieve state-of-the-art results with limited data.

Contribution

It presents a comprehensive multi-parameter evaluation and optimization of CRNN OCR models, demonstrating high accuracy with significantly less training data.

Findings

Achieved 0.44% CER with only 10,000 lines of training data

Identified key factors influencing OCR performance such as binarization and network architecture

Provided ablation studies on training pipeline components

Abstract

We investigate how to train a high quality optical character recognition (OCR) model for difficult historical typefaces on degraded paper. Through extensive grid searches, we obtain a neural network architecture and a set of optimal data augmentation settings. We discuss the influence of factors such as binarization, input line height, network width, network depth, and other network training parameters such as dropout. Implementing these findings into a practical model, we are able to obtain a 0.44% character error rate (CER) model from only 10,000 lines of training data, outperforming currently available pretrained models that were trained on more than 20 times the amount of data. We show ablations for all components of our training pipeline, which relies on the open source framework Calamari.