Analisis Perbandingan Evaluasi Deep Learning Untuk Klasifikasi Gaya Arsitektur

Abstract

This study compares the performance of several deep learning models in architectural style classification using architectural image datasets. The models used include InceptionResNetV2, VGG16, MobileNetV2, and ResNet50V2. The data is processed through image augmentation techniques to improve model generalization. Evaluation is carried out using accuracy, precision, recall, F1-score, and Confusion Matrix metrics to measure the effectiveness of the classification. The results show that InceptionResNetV2 and ResNet50V2 have the best performance with an accuracy of 84%, followed by MobileNetV2 (79%) and VGG16 (71%). More complex models show better ability in capturing visual patterns than lighter models. The results indicate that the use of deeper deep learning models can improve the accuracy of architectural classification. This research is expected to contribute to the development of more accurate and efficient architectural classification systems for various applications, including cultural conservation and architectural design.