Beef Image Classification Using the Inception V3 Transfer Learning Model

Economic conditions and increased incomes can impact rising prices for beef and other consumer products. The increase in consumption needs in the community causes the high cost of beef in the market. The high cost of meat is often taken advantage of by irresponsible people by selling beef that is not fresh or mixing fresh and not-fresh meat. The level of freshness of the beef itself can be seen manually by looking directly at the color and texture of the beef. However, not all people can distinguish which meat is fresh and which is not. A model that can classify beef automatically is needed. This research will use machine learning to predict fresh, half-fresh, and spoiled meat images. The machine learning model will use the Inception V3 transfer learning architecture model. The research contribution is using additional datasets to produce different results and obtain the beef image classification model. The results of the model performance evaluation show an accuracy value of 87.14%, a precision of 87.14%, a recall of 87.14%, and an f1 score of 88.28%. These results indicate that the Inception V3 architectural model is the best fit for use in beef classification. The information obtained from the experiments shows that the model evaluation results using Inception are more optimal than the Xception evaluation results from earlier research.