Maturity Detection of Crystal Guava Fruit UsingConvolutional Neural Network Algorithm

Abstract

Guava fruit in Latin is called Psidium Guajava L. is a tropical plant originating from Brazil, and distributed to Indonesia. One of the
fruit commodities found in Indonesia which is a leading commodity and continues to increase in production is guava. Guava is a
climacteric fruit due to chemical changes, namely the activity of the enzyme pyruvate which causes an increase in the amount of
acetaldehyde and ethanol so that CO2 production increases and ethylene produced during fruit ripening will increase the respiration
process. This research focuses on crystal guava fruit which is very commonly found in Indonesia, especially in East Java Province.
Where at the time of harvesting guava fruit of course have a different maturity. From these problems, researchers used the
Convolutional Neural Network (CNN) Algorithm to identify ripe guava fruit (Psidium Guajava L.) with image classification. From
this study, the CNN model using the VGG16 architecture that has been created in this study has an accuracy of 96% which is
approximately the same as the comparison of the accuracy of other CNN models and gets good performance on the testing model with
an accuracy rate of 83%.