Classification of cendrawasih birds using convolutional neural network (CNN) keras recognition

Abstract

Classification is part of predictive modeling and supervised learning. This method is used to determine the data class based on the previous value. In solving certain cases, there are various classification methods with varying degrees of accuracy. Convolutional Neural Network (CNN) is part of the Multilayer Perceptron (MLP) for processing two-dimensional data. CNN is also part of the Deep Neural Network and is applied to image objects. From several sources, it is stated that the classification process using images is not properly implemented in this MLP. Of course, this will result in the accuracy of the method in handling certain cases. In this study, the object classification process uses hard recognition to determine the accuracy value of the method using the object of the bird of paradise. From the results of this study, a training model was conducted using 10 ephocs with an accuracy value of 0.0850 while a loss value of 2.5658. So these results indicate that MLP can successfully complete the classification process using images.