Implementasi YOLOv10 untuk Deteksi Makanan Sebagai Pencatatan Asupan Gizi Protein Secara Realtime

Abstract

Nutrition refers to substances derived from food that are essential for the body to grow, develop, and maintain normal organ function. Protein is one of the primary components of macronutrients and plays a vital role in maintaining human health. It functions to build and repair body tissues, as well as in the formation of enzymes, hormones, and antibodies. An imbalance in nutritional intake, including insufficient or excessive protein, can lead to various health problems such as malnutrition, obesity, and metabolic disorders. Nutritional sources are food materials that contain the necessary nutrients required by the body for optimal growth, maintenance, and functioning. Animal-based foods are major sources of protein, such as eggs, milk, meat, poultry, fish, and shellfish. Plant-based protein sources include soybeans and their processed products, such as tempeh and tofu. Each individual has different protein requirements depending on factors such as age, body weight, and height. A lack of nutritional knowledge among parents and children is one of the contributing factors to unbalanced protein intake. Therefore, it is important to conduct research that can accurately assess protein nutritional intake. This study utilizes YOLOv10 to detect the protein content in seven types of food: chicken, tofu, tempeh, eggs, catfish, tuna, and beef jerky. The proposed method yielded an accuracy of 74.79% using a dataset consisting of 2,770 food images.