Screening Child’s Screen Time

Abstract

The Child iGuardian (CiG) is a mobile application to monitor the child’s screen-time and physical activities. The speciality of CiG is that users can capture their children’s emotional changes, sleepiness, web browsing behaviors, mouse movements, keyboard patterns when using digital devices and also assess physical activities. We used CNN, LSTM, XG Boost, Multinomial Naive Bayes, FFNN, SVM and Multiple Linear Regression algorithms to implement the models. For the facial expression and sleepiness recognition models, we used CNN and LSTM algorithms with 80% and 82% accuracies respectively. Web tracking and relationship between academic performance and web usage model implemented using Multinomial Naive Bayes and Decision Tree algorithms with 94% and 82% accuracies. XG Boost with 81% and FFNN with 97% accuracies are used to implement mouse movement and key pattern recognition models. Physical activity recognition and sleepiness analysis models are implemented using Logistic Regression and Multiple Linear Regression algorithms with 98% and 96% accuracies respectively.