Malnutrition Detection using Deep Learning Models

The enormous worldwide health problem of malnutrition affects millions of people from many cultures and age groups. Early detection and response are crucial to averting major health effects and improving patient outcomes. The goal of this research is to create two ways of identifying malnutrition. The first method analyses photos using the EfficientNetB4 model to evaluate malnutrition status directly from images, while the second method applies the ResNet34 model to estimate Body Mass Index (BMI) from visuals. The first method uses the most advanced classification model, EfficientNet-B4, to directly classify images to determine whether or not a person is malnourished. The model is created utilising a variety of picture datasets that show alterations in body composition, visual indications of malnutrition, and changes in body form. After thorough training and validation, the EfficientNet-B4 model can identify malnutrition with high accuracy using only photo inputs. The second approach concentrates on calculating BMI, a standard indication for determining malnutrition. To analyse photos and calculate the BMI of the people pictured, we use the ResNet34 model, a popular convolutional neural network. We seek to create an accurate system that can predict BMI from photographs without the need for direct measurements by training the model on a dataset of labelled images with matching BMI values.