Fruit Ripeness Based on RGB, HSV, HSL, L*a*b* Color Feature Using SVM

In this study, we have created a fruit ripeness dataset for 8 categories, namely Ripe Mango, Ripe Tomato, Ripe Orange, Ripe Apple, Unripe Mango, Unripe Tomato, Unripe Orange, and Unripe Apple. Based on the fruit ripeness dataset, we build a classification model of fruit ripeness using the SVM algorithm. Color feature extraction implemented in this study is RGB, HSV, HSL, and L * a * b *. To determine fruit ripeness, we done by predict image input to the model generated. Based on the experiment result, we have found that the best SVM model in determining fruit ripeness is the 6
^th
degree polynomial kernel and by extracting HSV color features. We evaluated the model generated based on the value of accuracy, precision, recall, and F-Measure. The best performance of our system for accuracy, precision, recall, and F-Measures are 0.76, 0.80, 0.76, and 0.78, respectively.