Application of Convolutional Neural Network Method with MobileNet V1 and ResNet-152 V2 Architecture in Batik Motif Classification

会议论文
发表日期：

2024年

查看全文

Aulia Chusnyriani Sani Zulkarnaen,
I Gusti Ngurah Rejski Ariantara Putra,
Nada Fauzia Reviana,
Rahmawati Hidayah,
Nur Ibrahim,
Nor Kumalasari Caecar Pratiwi,
Yunendah Nur Fuadah

摘要

Indonesia is a country that has diverse natural resources, cultures, and languages. One of the cultural diversities in Indonesia is Batik, which is an Indonesian cultural heritage consisting of cloth drawn by hand using traditional techniques. To assist the public in recognizing various batik motifs, a classification method was developed to identify the types of batik through input images. The classification method uses Convolutional Neural Network (CNN) based on MobileNet V1 and ResNet-152 V2 architecture. This research uses a dataset consisting of 3300 batik images from six different batik motifs, namely ceplok, parang, nitik, megamendung, kawung, and tambal. The optimal classification model was obtained using ResNet-152 V2 architecture with shear pre-processing method and RMSprop optimizer with test accuracy value of 89.67% and validation loss of 0.44.