Child Emotion Recognition via Custom Lightweight CNN Architecture

Advancements in e-learning coupled with the recent pandemic has resulted a paradigm shift when it comes to remote-based education. Whilst this its benefits most important being continuation of curriculum there are various risks posed younger learners. Their exposure wide variety content including that malicious intent poses serious risk emotional well children and is concern for parents. This chapter presents online child emotion recognition framework enabling parents monitor their childs well-being whilst engaging content. The result breakthroughs deep learning facilitating development lightweight convolutional neural networks can be deployed without need specific hardware requiring Graphical Processing Units. also covers security concerns presenting an offline rather than cloud-based inference mechanism.