In this paper we introduce a method that significantly reduces the character
error rates for OCR text obtained from OCRopus models trained on early printed
books. The method uses a combination of cross fold training and confidence
based voting. After allocating the available ground truth in different subsets
several training processes are performed, each resulting in a specific OCR
model. The OCR text generated by these models then gets voted to determine the
final output by taking the recognized characters, their alternatives, and the
confidence values assigned to each character into consideration. Experiments on
seven early printed books show that the proposed method outperforms the
standard approach considerably by reducing the amount of errors by up to 50%
and more.
Description
Improving OCR Accuracy on Early Printed Books by utilizing Cross Fold
Training and Voting
%0 Generic
%1 reul2017ocrvoting
%A Reul, Christian
%A Springmann, Uwe
%A Wick, Christoph
%A Puppe, Frank
%D 2017
%K das_2018_1 dnn myown
%T Improving OCR Accuracy on Early Printed Books by utilizing Cross Fold
Training and Voting
%U http://arxiv.org/abs/1711.09670
%X In this paper we introduce a method that significantly reduces the character
error rates for OCR text obtained from OCRopus models trained on early printed
books. The method uses a combination of cross fold training and confidence
based voting. After allocating the available ground truth in different subsets
several training processes are performed, each resulting in a specific OCR
model. The OCR text generated by these models then gets voted to determine the
final output by taking the recognized characters, their alternatives, and the
confidence values assigned to each character into consideration. Experiments on
seven early printed books show that the proposed method outperforms the
standard approach considerably by reducing the amount of errors by up to 50%
and more.
@misc{reul2017ocrvoting,
abstract = {In this paper we introduce a method that significantly reduces the character
error rates for OCR text obtained from OCRopus models trained on early printed
books. The method uses a combination of cross fold training and confidence
based voting. After allocating the available ground truth in different subsets
several training processes are performed, each resulting in a specific OCR
model. The OCR text generated by these models then gets voted to determine the
final output by taking the recognized characters, their alternatives, and the
confidence values assigned to each character into consideration. Experiments on
seven early printed books show that the proposed method outperforms the
standard approach considerably by reducing the amount of errors by up to 50%
and more.},
added-at = {2017-12-18T09:43:01.000+0100},
author = {Reul, Christian and Springmann, Uwe and Wick, Christoph and Puppe, Frank},
biburl = {https://www.bibsonomy.org/bibtex/2a693c55d96790a6c464dcdda6fb2a373/chwick},
description = {Improving OCR Accuracy on Early Printed Books by utilizing Cross Fold
Training and Voting},
interhash = {b84f57bc0d60ecd17252b86bdfcf07bf},
intrahash = {a693c55d96790a6c464dcdda6fb2a373},
keywords = {das_2018_1 dnn myown},
note = {cite arxiv:1711.09670},
timestamp = {2017-12-18T09:43:01.000+0100},
title = {Improving OCR Accuracy on Early Printed Books by utilizing Cross Fold
Training and Voting},
url = {http://arxiv.org/abs/1711.09670},
year = 2017
}