End-to-End Approach for Recognition of Historical Digit Strings

TL;DR

This paper introduces an end-to-end deep learning method for recognizing handwritten date strings in historical documents, achieving high accuracy without segmentation or heuristic steps, and outperforming existing models.

Contribution

The paper presents a novel segmentation-free deep learning approach using a modified VGG-16 model for recognizing handwritten date strings in historical documents.

Findings

Achieved 93.2% recognition accuracy on ARDIS dataset.

Outperformed the CRNN model in handwriting recognition tasks.

Demonstrated effectiveness without segmentation or heuristic methods.

Abstract

The plethora of digitalised historical document datasets released in recent years has rekindled interest in advancing the field of handwriting pattern recognition. In the same vein, a recently published data set, known as ARDIS, presents handwritten digits manually cropped from 15.000 scanned documents of Swedish church books and exhibiting various handwriting styles. To this end, we propose an end-to-end segmentation-free deep learning approach to handle this challenging ancient handwriting style of dates present in the ARDIS dataset (4-digits long strings). We show that with slight modifications in the VGG-16 deep model, the framework can achieve a recognition rate of 93.2%, resulting in a feasible solution free of heuristic methods, segmentation, and fusion methods. Moreover, the proposed approach outperforms the well-known CRNN method (a model widely applied in handwriting recognition tasks).