<![CDATA[Palembang jumputan fabric is one of Indonesia's cultural heritages that is unique in its motifs and manufacturing techniques. However, the lack of public understanding of the meaning of motifs and competition with other traditional fabrics are challenges in its preservation. This research aims to develop a classification model of Palembang jumputan fabric motifs using the Convolutional Neural Network method with MobileNetV2 architecture and transfer learning approach. The dataset used consists of 800 images of four types of motifs, namely Bintik Tujuh, Pola, Tabur, and Terong. The data is divided into 80% training, 10% validation, and 10% testing. The model was trained using four types of optimisers, namely AdamW, Adagrad, Nadam, and SGD, with training parameters of 100 epochs, batch size 32, and learning rate 0.001. The test results showed that AdamW gave the highest accuracy of 97%, followed by Nadam 96%, Adagrad 95%, and SGD 90%. The model recognised the motifs well, especially the Bintik Tujuh and Tabur motifs which achieved 100% accuracy. With these results, artificial intelligence can be utilised to support the preservation of Palembang jumputan fabrics through motif recognition technology.]]>
Copyrights © 2025
