Nutrient Detection in Complementary Feeding Using Convolutional Neural Network Algorithm

Infants or toddlers are an age group that is vulnerable to nutritional problems. One of the problems related to nutrition that often occurs is stunting. Stunting is a health problem because it is associated with the risk of morbidity and mortality. And one of the factors causing stunting in infants or toddlers is due to the lack of parental knowledge of the nutritional content of complementary foods. Therefore, effective preventive measures are needed to overcome this problem. This project builds a detection system by developing a Machine Learning model using Convolutional Neural Network (CNN) algorithm with MobileNetV2 architecture and interface development using java programming language, to inform the nutritional content of complementary food ingredients. This detection process is carried out in two stages: first using machine learning to detect images of complementary food ingredients and second detecting nutrients by matching the complementary food ingredients generated from machine learning with the firebase database, where the database contains nutrient content taken from the https://nilaigizi.com/ website. The dataset used consists of 11 labels of commonly used basic complementary food ingredients, which are obtained from the Kaggle website and obtained through the scrapping method from the internet. From the use of these datasets, the accuracy result is 97%, and the validation accuracy is 94%. And the detection system built can display nutritional detection results that match the input object. The results of this project are expected to support the development of nutritional science on complementary food ingredients.