Rice Quality Classification Model Using Machine Learning Technique

This study proposes a rice quality classification method using Convolutional Neural Network (CNN) to improve the accuracy and efficiency in classifying rice into premium, medium, and low. This study involves processing digital images of rice with data augmentation techniques, feature extraction using convolution layers, and classification using CNN models. The dataset used consists of rice images obtained through direct photography and open sources. This data is then divided into training, validation, and test data to improve model performance. Model training was carried out using Google Colab with Adam optimization and ReLU activation function. The test results showed that the CNN model was able to classify with an accuracy level of 85%, higher than conventional methods. Rice grain image data was collected and processed through preprocessing stages such as normalization and segmentation, then feature extraction was carried out. The rice image feature extraction values obtained include: Red 164.7719; Green 68.3355; Blue 61.7290; Hue 0.4037; Saturation 0.6625; Value 0.6462; and Area 493023 pixels.