Klasifikasi Kualitas Biji Jagung Menggunakan Metode Faster Region Convolutional Neural Network

Abstract

Maize is the second food commodity after rice. Corn also has various qualities. Determining the quality of corn kernels plays an important role in determining the selling price of corn kernels, in determining the quality of corn kernels still uses manual labor which causes subjectivity in the assessment, so that the assessment of the quality of corn kernels is not fully based on the object of the corn kernels and will cause differences in opinion between observers and other observers. The quality of corn kernels can be seen physically based on the texture of the corn kernels, which is seen from the crunchiness of the germ and the color of the corn kernels. This research produces a system that can classify the quality of corn kernels by looking at the color, texture and shape of the germ of corn kernels to minimize subjectivity in determining the quality of corn kernels. The quality of corn is then divided into five levels, namely Quality A, Quality B, Quality C, Quality D and Quality E. This research uses the Faster Region Convolutional Neural Network algorithm and uses ResNet50 as feature extraction. The data used in this study amounted to 2500 data which were then divided into 2000 training data, 250 validation data and 250 testing data. After testing, this research resulted in an accuracy of 95.2%.