Off-line Handwritten Textline Recognition Using a Mixture of Natural and Synthetic Training Data.
Tamás VargaHorst BunkePublished in: ICPR (2) (2004)
Keyphrases
- training data
- real world
- domain knowledge
- character recognition
- decision trees
- data sets
- supervised learning
- training process
- test data
- learning algorithm
- classification accuracy
- mixture model
- prior knowledge
- training examples
- training set
- learned from training data
- training samples
- real images are presented
- word spotting
- character segmentation
- training dataset
- gaussian distribution
- noisy data
- class labels
- neural network