Classification of Meat Freshness using Deep Learning

Abstract

Meat is a source of food protein that is quite popular among people. The price of meat tends to be expensive, causing many meat sellers to mix fresh meat with less fresh meat. Meat that is not fresh enough can affect the health of people who consume it. Based on these problems, research was created that can differentiate the freshness of meat using digital images. In this research, the meat that will be used as research objects is beef and chicken. The freshness of meat will be differentiated based on color and texture using deep learning. The processes carried out are the preprocessing process, data augmentation, feature extraction is carried out using color features and texture features using the Gray Level Cooccurrence Matrix (GLCM) method, training process and classification process using deep learning. The training process is carried out using the CNN method, such as ResNet 18, ResNet 34 and ResNet 50 architectures. The test results using three types of ResNet architecture obtained the best results using ResNet18, epoch 10 with an accuracy of 94%.