Klasifikasi Kain Ulos Simalungun Secara Real-Time Menggunakan SSD MobileNetV2 FPN Berbasis Android

Abstract

Ulos cloth is one of the Simalungun Batak cultural heritages that has symbolic meaning in various traditional ceremonies. The lack of literature, comprehensive access to Simalungun Ulos fabrics, and the lack of introduction to the younger generation are common problems in the sustainability of local culture that is passed down. The main issue faced is the limited knowledge of the various types of Simalungun Ulos fabrics, making it difficult for knowledge to be passed on to the younger generation as the pillars of cultural succession. Simalungun ulos fabric recognition can be done with an image classification-based approach. This research aims to implement the MobileNetV2 SSD method with a combination of FPN (Feature Pyramid Network) which is integrated into mobile android. The final results showed that the model got an mAP value of 95.03% with an IoU range of 0.5-0.95. This application is 216MB in size which includes descriptions, uses, how to use, meanings and additional explanations for 10 types of Simalungun Ulos fabrics. Based on the technical testing and feasibility testing obtained, it can be concluded that the system built with SSD MobileNetV2 FPN is good at classifying in real-time 10 types of Ulos Simalungun fabrics and can be well received by the younger generation as a medium of cultural recognition and education.