Auto-ML Deep Learning for Rashi Scripts OCR
This solves the problem of digitizing religious Jewish literature for scholars and archivists, but it is incremental as it applies existing deep learning methods to a specific domain.
The authors tackled OCR for manuscripts printed in the ancient Rashi Hebrew font by proposing a scheme using CNN and LSTM, achieving over 99.8% accuracy on a dataset of more than 3 million annotated letters.
In this work we propose an OCR scheme for manuscripts printed in Rashi font that is an ancient Hebrew font and corresponding dialect used in religious Jewish literature, for more than 600 years. The proposed scheme utilizes a convolution neural network (CNN) for visual inference and Long-Short Term Memory (LSTM) to learn the Rashi scripts dialect. In particular, we derive an AutoML scheme to optimize the CNN architecture, and a book-specific CNN training to improve the OCR accuracy. The proposed scheme achieved an accuracy of more than 99.8% using a dataset of more than 3M annotated letters from the Responsa Project dataset.