Klasifikasi Bentuk Bingkai (Frame) Kacamata Menggunakan CNN dengan Arsitektur Inception V3 dan Augmented Reality Berbasis Android

Abstract

Glasses are not only a type of vision aid for people with eye diseases, but they are also an increasingly popular part of the fashion world. The choice of eyeglass frame design can influence a person's appearance in clothing, so when making a choice you must pay attention to two important aspects, namely style and comfort, and can change the impression on a person's face. When designing eyeglass frames, it is necessary to use the science of measuring the human body, because each human organ's size and shape are different from each other. So, with the diversity of human facial shapes, it becomes very important in making the choice of eyeglass frames and the challenge in conducting research to build an application to recommend eyeglass frames according to face shape. With the current technological era, it is possible to apply Artificial Intelligence (AI) and Machine Learning (ML) to be the best solution to answer these challenges. Several studies have tried to classify facial shape using ML, with the best results using the Inception V3 architecture. In this research, a Unity 3D-based application was developed that combines Augmented Reality (AR) with ML to recommend eyeglass frame shapes based on face shape. Inception V3 model training results show performance improvements over time. However, it is necessary to overcome overfitting in validation data. In testing the test data, the model achieved an accuracy of around 78.6%, indicating good prediction ability. This technology has the potential to help consumers make more informed decisions when selecting glasses