Identifikasi Kematangan pada Buah Mangrove Menggunakan Metode Faster Region Convolutional Neural Network pada Platform Mobile

Abstract

Most of the mangrove farmers believe that the mangrove fruit they want to plant in the rehabilitation case already has a maturity level, but the yield after replanting is not suitable. Therefore we need an application to make it easier to recognize mangrove fruit based on the level of maturity according to its properties. The purpose of this study was to identify the maturity level of mangrove fruit and divide the results of the ripeness into three digital images, namely raw, half-ripe and near-ripe through image processing in RealTime. This study uses the Faster Region Convolutional Neural Network method with an accuracy of 98.4%. The identification test was carried out as many as 6 experiments with the lowest loss results in the 2nd experiment with 1% loss results, 98% Precision and 96% Recall for mangrove fruit identification level data with a total test data of 450 images