Improving OCR Accuracy on Early Printed Books by Utilizing Cross Fold Training and Voting.
Christian ReulUwe SpringmannChristoph WickFrank PuppePublished in: DAS (2018)
Keyphrases
- optical character recognition
- high accuracy
- post processing
- prediction accuracy
- test set
- preprocessing
- training speed
- voting method
- scanned documents
- error correction
- classification accuracy
- training set
- error rate
- online learning
- random selection
- active learning
- digital documents
- mathematical expressions
- information retrieval
- text recognition
- printed text