Implementation of Convolutional Neural Networks for Batik Image Dataset

Vina Ayumi (Universitas Mercu Buana)
Ida Nurhaida (Universitas Mercu Buana)
Handrie Noprisson (Universi tas Mercu Buana)

International Journal of Computing Science and Applied Mathematics-IJCSAM
Vol. 8 No. 1 (2022)

One method of image recognition that can be used is a convolutional neural network (CNN). However, the training model of CNN is not an easy thing; it takes tuning parameters that take a long time in the training process. This research will do Batik pattern recognition by using CNN. From the experiment that we conducted, the result shows that the feature extraction, selection, and reduction give the accuracy more significant than raw image dataset. The feature selection and reduction also can improve the execution time. Parameters value that gave best accuracy are: epoch = 200, batch_size = 20, optimizer = adam, learning_rate = 0.01, network weight initialization = lecun_uniform, neuron activation function = linear.