Handwritten Javanese script recognition method based 12-layers deep convolutional neural network and data augmentation

Although numerous studies have been conducted on handwritten recognition, there is little and non-optimal research Javanese script recognition due to its limitation basic characters. Therefore, this proposes the design of a Script method based twelve layers deep convolutional neural network (DCNN), consisting four convolutions, two pooling, five fully connected (FC) layers, with SoftMax classifiers. Five FC were proposed in conduct learning process stages achieve better outcomes. Due limited number images dataset, an augmentation needed improve performance. This obtained 99.65% accuracy using seven types geometric DCNN model for 120 character classes. It consists 20 characters plus 100 others from compound vowels