Binarization-free OCR for historical documents using LSTM networks.
Mohammad Reza YousefiMohammad Reza SoheiliThomas M. BreuelEhsanollah KabirDidier StrickerPublished in: ICDAR (2015)
Keyphrases
- document images
- historical documents
- optical character recognition
- handwriting recognition
- document analysis
- document image analysis
- document processing
- historical manuscripts
- handwritten document images
- printed documents
- character recognition
- character segmentation
- preprocessing
- text recognition
- recurrent neural networks
- handwritten documents
- machine learning
- image collections
- color images
- computer vision
- document image retrieval