<![CDATA[Food fraud has emerged as a significant global issue, threatening public health, economic stability, and consumer trust across the food supply chain. In the context of rice—a staple consumed by more than half of the world’s population—the proliferation of counterfeit products poses a critical risk. This study aims to develop a deep learning-based classification model using Convolutional Neural Networks (CNN) to accurately distinguish between medium-grade, premium, and counterfeit rice. The research involved the systematic collection of 100 grain images per rice category, followed by preprocessing, data augmentation, and model training using an optimized CNN architecture for image-based classification. The dataset was split into training, validation, and testing subsets with a 60:20:20 ratio. The model was trained over 12 epochs, achieving a training accuracy of 95%. Evaluation using the test set yielded identical accuracy, with the confusion matrix confirming perfect classification across categories. External validation further demonstrated the model’s robustness and generalizability. The findings highlight CNN’s potential as an effective tool for enhancing food safety monitoring systems and combating rice fraud. This AI-driven approach contributes to agricultural quality control and emphasizes the role of machine learning in promoting food security and authenticity assurance. However, CNN models face limitations, including susceptibility to overfitting when trained on insufficiently diverse data and high computational demands during training. These challenges underscore the need for diversified datasets and the exploration of alternative architectures offering comparable performance with greater computational efficiency.]]>
