MEAT IMAGE CLASSIFICATION USING DEEP LEARNING WITH RESNET152V2 ARCHITECTURE

Abstract

Meat is a food ingredient that can be consumed by humans and consists of essential nutrients, especially protein, which are needed for various physiological functions in the human body. Beef, goat and pork are meats that are commonly used by Indonesian people as daily processed foods. A very high level of meat consumption results in a high economic value of meat consumption. However, many people do not know how to distinguish between the types of beef, mutton and pork. This study aims to classify types of beef, goat and pork using the ResNet152V2 algorithm. The data used are 600 images with 200 images of beef, 200 images of mutton and 200 images of pork. The process carried out is pre-processing using 4 stages, namely image augmentation, image sharpness process, then the image is resized to adjust the size needed by the algorithm. The last pre-processing is to perform the image normalization process. After the pre-processing is done, then the data training stage is carried out using the ResNet152V2 algorithm to build a classification model and then the model is tested against data testing to get the results of the optimal classification of pork, goat and beef images by looking at the results of accuracy and loss values.