Accuracy potential of the Convolutional Neural Network (CNN) in recognizing traditional clothing

Abstract

The diversity of cultures in Indonesia is proof that Indonesia is a country that is rich in cultural diversity. Many foreign tourists who want to know about culture in Indonesia are not directly proportional to the media to introduce culture in Indonesia. Therefore, this study aims to classify images of traditional clothing by detecting images of traditional clothing sent to the application to determine the name of the traditional soldier. These images will be converted into vectors and processed to find the closest similarity level. The Deep Learning method which currently has the most significant results in image recognition is the Convolutional Neural Network (CNN). The analysis carried out resulted in an accuracy of 0.7934 with an epoch of 20 and a data set of 700 data. The accuracy value is 0.7934 which is a large enough number to determine the correct classification of image objects. This is proven by testing on 10 different images and only 1 image is inaccurate with 90% accuracy.