Creation of Dataset for Shan Handwritten Alphabet Recognition System of Myanmar

Nowadays, the handwritten alphabet recognition system plays an important role in any language (English, Hindi, Tamil, Chinese, Myanmar, etc.). There is limited research on the Myanmar handwritten alphabet recognition system, especially on Shan State alphabet recognition. One prevalent problem in this area of research is that there is neither a standard dataset for the Shan alphabet nor any research on the Shan handwritten alphabet recognition system. To overcome this problem, this study aims to create an image dataset for Shan alphabet handwritten recognition. The dataset developed consists of 19 alphabet classes that are widely used to write Shan text in Shan state of Myanmar due to the distinctive orthographic representation of the Shan alphabets. This dataset was collected by 1000 Shan native participants of different age groups in Shan state. Thus, a total of 19,000 images were created for each class in this dataset. This dataset will be useful for Shant state alphabet recognition system in Myanmar and assist in natural language processing (NLP) based applications in Myanmar.