Character recognition of handwritten alphabets from Telugu language using Machine Learning

Machine Translation and Natural Language Processing are two key fields of technology that are bringing the world together. Machine Translation eliminates the barrier for communicating with people who speak a different tongue. The roots of Machine Translation are identifying the language of the word and then translating it into the target language. Identifying the language of the word can be easier if each letter in the word can be identified. This brings us to “Character Recognition†. This thesis focuses on reviewing the existing methods for Character recognition of handwritten characters using Machine Learning and applying those methods to recognize handwritten characters of alphabets from the language of Telugu. The interesting aspect of Machine Learning is that the machines which are trained to translate, say Chinese, can be trained on a different language like Hindi or Telugu. This is a very powerful feature and the one that thesis tries to explore. The current state of technology has gone beyond recognizing the characters and can also translate those characters across multiple languages. The machine learning model developed during this thesis can identify the alphabets of Telugu language with an accuracy of 70 – 90% depending on the character.