Artificial Intelligence-Based Banana Ripeness Detection

Abstract:

This paper aims to construct an artificial intelligence model that detects the level of ripeness of bananas according to three categories of the Von Loesecke scale: totally green, yellow with green tips and yellow with brown spots. The manual methods typically used to classify the degree of ripening of bananas are unreliable since they are very subjective; therefore, it is necessary to automate this procedure. The features leveraged by our model are extracted using the HSV (Hue, Saturation, Value) color model obtained from the images. The Multiclass Support Vector Machine (SVM) model based on OVO (one-vs-one) with an Radial Basis Function (RBF) kernel function is implemented as the proposed classifier. The used dataset comprises 1242 images divided into 80% for training and 20% for test data. The accuracy obtained using the proposed model was 98.89%, outperfoming the state-of-the-art methods (We have published the source code at https://github.com/Jorge260399/AI-bases-framework.).

Año de publicación:

2023

Keywords:

• Artificial Intelligence
• Banana ripeness
• SVM multiclass
• RBF
• HSV

Fuente:

scopus