Sinhala Handwritten Character Recognition using Convolutional Neural Network

摘要

Handwritten character recognition is widely used for the English language. It is difficult to create a character recognition model for south Asian languages because of its shape and compound characters. Among other South Asian languages (e.g.: - Tamil, Hindi, Malayalam, etc.) Sinhala characters are unique, because of their shape, which are having mostly curves and dots. These unique characteristics make it difficult to create a model to recognize Sinhala’s handwritten characters. Recognizing handwritten characters rather than typed characters is more complicated because the handwriting of each individual is varying from each other. Therefore the recognition of Sinhala handwritten character need to be improved. Convolutional Neural Network (CNN) is playing a vital role in character recognition by supporting the more efficient image classification. This research focuses on recognizing Sinhala handwritten characters using CNN. Google colaboratory platform is used for the experiment, and python programming language is used for the implementation part. In total, around 110,000 image data were used for the experiment. CNN’s performance was evaluated by training and testing the dataset by increasing the number of character classes. When it reaches 100 character class it shows reasonable accuracy of 90.27%. The model was trained by 5 sets of different 100 character classes. Finally, the overall accuracy of 82.33% is achieved for 434 characters. This model outerformed than similar systems.