Mendeteksi Salak BerLarva dan Tidak BerLarva Menggunakan Metode Convolutional Neural Network

Abstract

Salacca fruit is one of the cultivars of the salak pondoh (the young fruits with sweet taste), which has become a leading commodity in Sleman, the Special Region of Yogyakarta Province. The salak madu (honey salacca fruit ~Eng.) became known when it was identified for the first time in Sempu Sub-village (Baterante), Wonokerto Village, Turi District, Sleman Regency. The most prominent feature of this salak madu is the leaves are shorter when compared to other types of salak pondoh. The color of the fruit's skin is blackish-brown when it is young, and it gradually becomes shiny brown after getting old. The arrangement of the scales forms a line pattern. This research was initiated because some people were sometimes deceived by the salak madu that looked good even though it had larvae inside. Therefore, the researcher was willing to create a system to detect which salacca fruits had larvae inside and which did not, using the Convolutional Neural Network (CNN) method. Detecting the salacca fruits with larvae and with no larvae required an image of the salacca fruits with larvae and with no larvae; after that, the image that had been obtained would go through a training process to detect the salacca fruits, whether they have larvae or no larvae; and after that, the image would be tested for the accuracy. The test result obtained by this System was 80%.