Identifikasi Diabetes Melitus Tipe 2 Melalui Citra Lidah dengan Menggunakan Arsitektur Efficientnet Berbasis Android

Abstract

Diabetes Mellitus (DM) is a chronic, chronic disease characterized by blood glucose (blood sugar) levels exceeding normal Symptoms of DM can be observed in several parts of the body, including the tongue. The tongue can be one of the parameters in diagnosing several diseases, because the condition of the tongue can provide clues about the overall health condition. Some symptoms on the tongue that can be observed in people with diabetes mellitus include sores or white patches on the tongue and look drier. However, a comprehensive medical examination and other diagnostic tests are still required to ensure a proper diagnosis. EfficientNet architecture is an architecture from CNN algorithm which combines intelligent scaling techniques such as width, depth, and image resolution. The data used amounted to 500 data which was then divided into 400 training data and 100 testing data. The data goes through a pre processing stage such as resizing, cropping, flipping, and rotating image to 45 degrees and 90 degrees and then classified using the Convolutional Neural Network algorithm with the EfficientNet-B3 model architecture. Based on the results of data testing on the system carried out in this study, the results obtained in the form of a system that can classify the quality of eggs with an accuracy value of 97%. The model that has been created is then implemented into the Android application.