Klasifikasi Kualitas Buah Salak Menggunakan Algoritma Faster R-CNN Berbasis Android

Abstract

Salak with the scientific name Salacca zalacca is a type of plant classified in the Arecaceae tribe or palms with edible fruits. Salak plays an important role in increasing Indonesia's foreign exchange. In addition to being marketed in the local market, , salak fruit has also been exported to various countries in Asia. In 2018, salak exports reached 1,233 tons, 28% increase from the previous year. This number continued to increase in 2019 to 1,698 tons and made salak the fourth most exported fruit in Indonesia. The selection of the quality of salak fruit is very important and needs to be done carefully so that the distribution of salak fruit for long-distance delivery and export destinations can be carried out effectively. Producers and sellers of salak fruit must be able to produce and market good quality salak fruit so that it can satisfy consumer needs. The determination of the quality of salak fruit is still done manually with the sense of sight so this will require precision and will take longer. The existence of a difference perception of subjective quality assessment can cause obstacles in the salak fruit distribution process. In this research, there are 5 qualities of salak fruit that will be classified, namely good quality, semi-ripe, tear defects, dent defects and rotten defects using the Faster R-CNN algorithm. The study used 535 images which were divided into training data, data validation and data testing. After conducting the research, it can concluded that Faster R-CNN can classify 5 different classes with 92% accuracy value.