AN ANALYSIS OF BANANA RIPENESS LEVEL BY IMAGE USING RGB COLOR MODEL AND SUPERVISED LEARNING

Abstract

identification models of pixel color. There were two parts to the experiment: 1) Generate model of Identification of pixel color, which use three banana categories i.e., Cavendish Banana, Cultivated Banana, and Lady Finger Banana, which are two kinds are raw and ripe. All each kind uses 20 images as a total of 120 images. The experimental results of accuracy in the training part (decision tree, deep learning) are 90.90 and 98.62, and accuracy in the testing part is 93.57, 99.31. 2) Measuring ripeness level use three banana categories, which are two kinds are ripe and very ripe. All each kind uses 10 images as a total of 60 images. The pixel color identification model can measure ripeness levels for both ripe and very ripe bananas. The experimental results show that the Arithmetic Mean of Cavendish Banana, Lady Finger banana, and Cultivated banana are 94.40, 93.31, 81.04, respectively.