Deep Learning based Detection and Classification of Child Malnutrition

Malnutrition remains a pervasive global health issue, affecting millions of children worldwide and manifesting in various forms from undernutrition to overnutrition. This study addresses the critical need for early detection and intervention by developing a robust deep learning-based system for classifying children’s nutritional status. Utilizing a diverse dataset, the performance of a baseline Convolutional Neural Network (CNN) and various transfer learning models including ResNet, MobileNet, VGG16, and VGG19 was evaluated. The baseline CNN model, featuring five convolutional layers trained over 60 epochs, achieved a training accuracy of approximately 98.53% and a validation accuracy of $\mathbf{9 9. 8 3 \%}$. Among the transfer learning models, MobileNet and VGG16 exhibited superior performance, closely matching the baseline CNN’s effectiveness, while VGG19 demonstrated strong learning capabilities with slightly lower validation accuracy. ResNet, despite improvements in training accuracy, struggled with validation accuracy and generalization. This system has been integrated into a web application for real-time image classification, providing a valuable tool for healthcare professionals in the early detection and management of malnutrition. The findings in this study highlight the potential of deep learning models to enhance public health outcomes and support the global effort to combat child malnutrition through timely and accurate assessments.