Abstract: Recognition of handwritten digits is a interesting research topic in Optical Character Recognition (OCR) in recent years. In this work, a CRNN model is proposed which is the combination of convolutional neural network (CNN) and Recurrent neural network (RNN). In this model, GRU is used to replace the CNN fully connected layer part to achieve high recognition accuracy Firstly, CNN ensures that the best features of the original image can be extracted using multiple convolutional and pooling layers. Secondly, GRU is a fast recognition method, which established a sequential relationship between features in the hidden layer. Experiment result performed on Kannada MNIST handwritten digit dataset achieves 98.12% training accuracy and 98.95% validation accuracy and also achieved higher precision, recall and F1 score ratio.

Keywords: CNN, RNN, GRU and Kannada MNIST

| DOI: 10.17148/IARJSET.2021.81125