The color-texture features and machine learning approach for quality detection of coffee beans

The quality of coffee beans has an important effect on the price and consumer satisfaction; therefore, it is essential in the industry to determine the quality of coffee based on the beans. The coffee industry is a huge commodity, especially Robusta coffee, which is popular among many groups. However, not everyone can differentiate the quality of coffee beans consumed because this is only known by people involved in the coffee world. Accordingly, the method to classify the quality of Robusta coffee beans was developed using machine learning. The method consists of two main processes: feature extraction and classification. Feature extraction was applied based on color, shape, and texture features; then, classification was performed using several machine-learning algorithms. The performance evaluation used cross-validation with the K-Fold of 5 and 10. The highest accuracy value achieved was 96.11%. This value resulted from color and texture features utilizing k-nearest neighbor (KNN) and artificial neural network (ANN). The resulting performance shows that the proposed method accurately classified coffee beans.