Sistem Klasifikasi Motif Batik Berbasis EfficientNet-V2B0 dengan Rekomendasi Pemakaian Batik Menggunakan Metode Rule-Based Berdasarkan Jenis Acara

Oleh :
Muhammad Aditya Ikhsan Saputra - V3423054 - Sekolah Vokasi

Batik motif recognition still faces challenges in identifying motif types, understanding their philosophies, and determining the appropriate use of batik for different types of events. Therefore, this study aims to develop a web-based Batik AI System to assist in batik motif classification while providing information and recommendations for batik usage. The system applies Deep Learning technology with the EfficientNetV2B0 architecture to classify 11 batik motif classes. The development method uses the Prototype approach, consisting of data labeling, preprocessing, data augmentation, model training, and model evaluation. The system is developed by integrating the React frontend, FastAPI backend, PostgreSQL database, and AI model, while also applying a rule-based method to provide recommendations for batik usage.The evaluation results show that the EfficientNetV2B0 model achieved an Accuracy of 75.80%, Precision of 74.50%, Recall of 76.10%, and F1-Score of 73.90%. The Integration Testing results show that all system components can be integrated according to the designed workflow, while Black Box Testing shows that the main system functions operate as required. Therefore, the Batik AI System can assist in the process of batik motif recognition, the delivery of cultural information, and the provision of integrated recommendations for batik usage.