Title

MAIDE: A MOTHER’S AIDE IN RECOGNIZING AND ANALYZING BABY CRIES USING MACHINE LEARNING

Author(s)

Jessa Abadilla

Kylah M. Ostia

Affiliation

Misamis University

Misamis University

Contributor

Orong, Markdy Y. - Adviser

Caroro, Roseclaremath A. - Panel Chairman

Talirongan, Florence Jean B. - Panel Member

Talirongan, Hidear - Panel Member

Year

2024-11

Abstract

Recognizing and analyzing baby cries is crucial for identifying specific needs such as hunger, discomfort, belly pain, burping, or sleepiness. This study developed a tool, MAide, to address these challenges of recognizing baby cries and interpreting their meaning. The system integrated a real-time detection device that captures and processes baby cries, transmitting data to a mobile app. This device was developed using the Agile methodology and the Scrum framework, where modeling was used, Teachable Machine Learning, a platform built on TensorFlow, which implements a pre-trained Convolutional Neural Network (CNN) to classify and analyze baby cries based on a labeled dataset. The dataset utilized in this research comprises samples from Kaggle, GitHub, Splann, and Donate-a-Cry database. Nevertheless, the researchers had to examine these datasets closely to guarantee their accuracy and dependability. The MAide allows mothers to register devices and add their baby's information to detect baby cries and receive instant notifications about their baby's needs. This solution enhanced caregiving by enabling timely responses, reducing stress for new parents, and improving confidence in infant care. The system was evaluated and received high functionality, accuracy, and usability ratings. The MAide platform demonstrated that the IoT-enabled system significantly enhanced infant care by recognizing and analyzing baby cries, providing mothers with a reliable and efficient tool for understanding and promptly responding to their baby's needs.

Keywords

Baby Cry Recognition

Convolutional Neural Network

IoT

Teachable Machine Learning

TensorFlow