Beef Freshness Classification Using CNN with DCT and GLCM Feature Extraction

Abstract

The increasing global demand for beef, which has risen by 13.9% over the past decade, underscores the growing importance of ensuring meat quality and freshness in the food industry. Conventional methods for assessing beef freshness rely on manual visual inspection, which is time-consuming, subjective, and often inaccurate. To address these limitations, this study proposes a hybrid approach that integrates the Discrete Cosine Transform (DCT), Gray Level Co-occurrence Matrix (GLCM), and Convolutional Neural Network (CNN) techniques for automated beef freshness classification. A dataset of fresh and spoiled beef images was used, followed by a series of preprocessing steps, feature extraction using DCT and GLCM, and classification through a CNN-based model. The integration of frequency-domain and texture-based features enhances the model’s ability to capture discriminative visual patterns associated with meat freshness. Experimental results demonstrate that the proposed model achieves an overall classification accuracy of 93%, with F1-scores of 0.94 for fresh meat and 0.93 for spoiled meat. These findings indicate that the DCT, GLCM, and CNN framework provides an efficient and reliable alternative to traditional inspection methods. The proposed approach contributes to the advancement of computer vision applications in food quality assessment, promoting improved automation, objectivity, and quality control across the meat supply chain.