Klasifikasi Jenis Buah Apel Secara Real-Time Menggunakan SSD-Mobilenet

Abstract

Apples are one type of fruit commodity that is very easy and often found in Indonesia. In classifying the types of apples, it is determined by several parameters, including weight, size, weight, characteristics, color, and many others. The grouping of apples in terms of shape and skin color is an important factor in the identification process. At the stage of identifying the type of apple manually by the human eye, the perception tends to be subjective due to color composition factors such as royal gala apples, fuji apples, and washington apples. Therefore, we need a tool with a system that can select apples based on the type. This real-time apple type classifier will use the SSD-MobileNet method which is used as feature extraction and the SSD model will be used as object detection in the image. This study resulted in an accuracy of 92.6% and for Royal Gala Apples 94.7%, Fuji Apples 92%, and Washington Apples 91.2%.