TS-Net: OCR Trained to Switch Between Text Transcription Styles.
Jan KohútMichal HradisPublished in: ICDAR (4) (2021)
Keyphrases
- text recognition
- printed documents
- handwriting recognition
- optical character recognition
- text extraction
- document processing
- page layout
- ocr systems
- document images
- document analysis
- information retrieval
- error correction
- high speed
- text lines
- scanned documents
- free text
- keywords
- handwritten text
- training set
- text retrieval
- character recognition
- document image analysis
- recognition errors
- text mining
- manually constructed
- speech recognition
- database programming
- text analysis
- text detection
- viterbi algorithm
- hybrid algorithm
- multilayer perceptron
- language independent
- neural network
- printed text