Palembang Tie-Dye Fabric Motif Detection Software Using The Yolov10 Method

Muhammad Firmansyah (a:1:{s:5:"en_US"s:33:"Universitas Multi Data Palembang "})
Tinaliah Tinaliah (Multi Data University Palembang)

International Journal of Health, Engineering and Technology
Vol. 5 No. 1 (2026): IJHET MAY 2026

Palembang's jumputan cloth shops face difficulties in manually identifying motifs, which can lead to inventory errors. This study aims to develop a jumputan cloth motif detection software using YOLOv10 for real-time inventory identification and management automation on Android. The research method combines Research and Development (R&D) with an experimental quantitative approach. The population consists of fabric stocks from three Palembang shops, a sample of 400 images (Titik Tujuh, Beras Tabur, Lereng, Keong) divided into training (80%), validation (10%), and testing (10%). Instruments include a smartphone camera, YOLOv10m, and Google Colab Pro; analysis uses precision, recall, mAP50-95, and confusion matrix. The results show mAP50-95 up to 99.50%, smartphone accuracy 90-100% (SGD is superior), user satisfaction 96.52% via USE Questionnaire, but decreases in low light (Keong 40%). Conclusion: The application effectively supports business efficiency and cultural preservation with a detection time of 3-5 seconds.