Deep Learning-based Identification of Rice Varieties: A CNN Approach

This research presents an imagebased classification algorithm for accurately identifying five distinct rice varieties: Arborio, Basmati, Ipsala, Jasmine, and Karacadag. Utilizing Convolutional Neural Networks (CNNs), the model effectively analyzes morphological, shape, and color features to differentiate between the varieties. The model was trained and evaluated on a large dataset of 75,000 images, achieving a high macro-average accuracy of 98%. Standard metrics such as F1-score, recall, and precision were used to assess the model's performance. This research demonstrates the potential of CNNs in automating rice variety classification, enabling improved quality control, efficient market segmentation, and optimized agricultural practices. By enhancing the accuracy and efficiency of rice classification, this technology contributes to the advancement of the agricultural sector and supports the sustainable production and distribution of this vital food source.