LAKASABrowsing by Author "Rocacorba, Arjay B."

Now showing 1 - 1 of 1

Ethnicity-inclusive malnutrition detection with convolutional neural network in the philippines
(Don Mariano Marcos Memorial State University - South La Union Campus, 2024-04) Cerezo, Angelo Gabriel S.; Perez, John Paul Rey B.; Rocacorba, Arjay B.; Soriben, Joy M.; Valdez, Azriel V.
In this study, the researchers developed a deep learning-based malnutrition detection model for public elementary schools in La Union and Pangasinan. Their primary objectives were to collect a suitable dataset, apply pre-processing techniques, train a Convolutional Neural Network (CNN) using AlexNet architecture, and evaluate its performance. They gathered 400 images from Bigbiga Elementary School and Sta. Rita West Elementary School, pre-processed, and augmented the data to create 1,200 training samples, 400 validation samples, and 80 test samples. Techniques like Horizontal Flip, Brightness Adjustments, Noise Reduction, and Random Rotation were used to enhance dataset quality. The CNN model trained on this data achieved a 66% test accuracy in identifying malnutrition. This research provides a foundation for early malnutrition detection, suggesting future improvements through additional pre-processing techniques, dataset expansion, and integrating the model into applications for broader lise, potentially improving public health outcomes