Handwritten Character Recognition using Neural Network

  • NayanaSajeev et. al


Abstract This paper is aimed at identifying the characters in a given scanned text and analysing the effects of modifying ANN's Models. Today Neural Networks are primarily used for pattern recognition tasks. The paper would identify the best approach to achieving more than 90% accuracy in the field of Handwritten Character Recognition (HCR). A lot of research has been done in the area of HCR, but it is still an open question as we still lack the highest precision. [1] describes the behaviours of various neural network models that are used in OCR. Neural Network is commonly used by OCR. In [2] Offline handwriting character recognition will be done through the Convolutionary Neural Network and TensorFlow. A procedure called Soft Max Regression is used to determine the probability of handwritten characters being one of many characters as it provides the values between 0 and 1 summing up to 1.