Child Monitoring System : Integrating Face, Emotion, and Activity Recognition for Enhanced Safety and Well-being

Abstract

This research work tackles the growing concern for child safety and security by developing a child monitoring system. The project aims to detect children, recognize their expressions, and identify both scheduled and spontaneous actions by analyzing CCTV footage. Current studies have used a variety of models, including Yolov5 and CNN, for identifying faces and emotions, as well as PCNN and HAR for identifying activities. These solutions, however, include models that are targeted towards adult emotions and do not precisely address the distinct emotional traits and behaviors of children. This study focuses on three detection models that are especially designed for children: face detection, emotion detection, and activity recognition. To address the drawbacks of existing datasets, a customized dataset has also been created for face, emotion, and activity recognition. Seven fundamental emotions of happy, sad, angry, disgust, surprise, fear, and neutral; as well as the two acts of crying and playing are included in the datasets for emotion and activity detection. This paper's major objective is to improve child security and safety through the implementation of a comprehensive child monitoring system. This technology gives parents peace of mind by accurately recognizing children, identifying their expressions, and detecting their actions. It provides children with the best protection and wellbeing possible due to its features, significantly improving their security and general well-being.