Machine learning with analysis-of-variance-based method for identifying rice varieties

Rice, which has the highest production and consumption rates worldwide, is one of the main nutrients in our country because it is economical and nutritious. Rice undergoes various stages of production from the field to the dinner tables, with the cleaning phase involving separation of rice from unwanted materials. There are different varieties of rice such as arborio, basmati, aracadaga, ipsala, and jasmine. Therefore, it is necessary to identify rice varieties based on customer demands. Rice variety identification is primarily based on its outer appearance, including color, size, and shape. In this study, 75,000 rice grain images, including 15,000 images for each variety, were considered. Using a combination of 12 morphological features, four shape features, and 90 color features obtained from five different color spaces, 106 features were extracted from the images. An analysis of variance (ANOVA) was employed to select high-rank features, which were then fed to a support vector machine (SVM) for classification. The experimental results demonstrated that the utilization of 40 high-rank features yielded impressive outcomes, with a validation accuracy of 99.89 %, cost value of 16, test accuracy of 99.81 %, and cost value of 28. These high accuracy rates are particularly noteworthy, as they indicate the robustness and reliability of the proposed methodology in distinguishing between different rice varieties with minimal error. This high level of precision is critical for practical applications, such as quality control in rice production, to ensure that customers receive the specific rice variety they demand. Implementing this methodology in real-world scenarios could significantly enhance the efficiency and accuracy of the rice variety identification processes, benefiting both producers and consumers.