A Roadmap on Handwritten Gujarati Digit Recognition using Machine Learning
Janardan Bharvad, Dweepna Garg, Shivam Ribadiya · 2021
One of the major difficulties in the area of pattern recognition and image processing is handwritten character recognition (HCR). For human, this seems to be very easy but for a machine, the task of recognizing handwritings is tedious. In case of machine, first the input needs to be scanned from file, image and real-time device like personal computers, digitizers and tablets. After that, the scanned input is translated in digital text in Handwritten Character Recognition process. There are two ways this can be done: online and offline. In online method, the input is taken at runtime whereas in offline method, the input is a scanned file. The mechanism where the machine understands an image of a handwritten script automatically is known as Offline handwritten character recognition. Various applications of HCR are processing of banks, mail sorting and reading of documents etc. As per the survey carried out, it was seen that the Offline handwriting is relatively challenging, to identify, as different people have different handwriting. As machine learning is a buzz word nowadays and most of the real-time applications are solved easily using the models of machine learning, hence our paper focuses on comparing various techniques of machine learning used to recognize the Gujarati handwriting digits. In Gujarat State, India, Gujarati script used by people to write the Gujarati language.