Deep Handwriting Recognition with Data Augmentation Techniques
Machine learning techniques have been successfully used in deciphering handwritten text. Deep learning techniques have made further improvements in this regard, however, they require substantial amounts of training data. This research aims to improve the effectiveness of machine learning techniques – deep learning algorithms in particular – on handwriting recognition. The main focus involves extending their ability to learn on smaller datasets. These improvements will enable handwriting recognition systems to be more capable of recognising written language where a limited amount of labelled data is available. These improvements allow for wider use of these systems across more regions, for greater accessibility, and for future related systems to be less reliant on the amount of data available. Therefore, the approach of the proposed research includes an investigation on image processing and machine learning, with a focus on the generation of more sample data from existing samples.
Supervisor: Dr Dane Brown