Development of Mobile Application for Thai Buddha Amulet Classification

Abstract

In this paper, we use transfer learning technique to train a custom Thai Buddha amulet coin classifier. The trained model is deployed on a server allowing a dedicated mobile application to request for class prediction results. The trained model achieves good accuracy using relatively small training image dataset. It is important to note that the primary objective of this work is not to evaluate the authenticity and quality of an amulet, but rather to create a prototype system that can be used as a tool to classify types of amulets.