Deep Learning-Based Multiclass Image Classifier for Emotion Identification

In an era marked by increasing mental health challenges, innovative solutions for real-time emotional monitoring are essential. This paper presents a non-invasive biometric system for emotion detection that utilises Convolutional Neural Networks (CNNs), a well-established deep learning architecture for image classification tasks, to identify feelings like happiness, sadness, anger, fear, surprise, and neutral emotions. Utilising the Facial Expression Recognition (FER) 2013 dataset, the research emphasises the significance of accurate emotion detection in healthcare, education, and workplace settings. The findings emphasise the system's potential to facilitate early interventions and support mental well-being, laying the groundwork for future advancements in emotional health monitoring. This research contributes to the ongoing development of more advanced emotion AI, aiming to capture the complexity of human emotions without oversimplification. It also addresses the outcomes and challenges encountered during implementation and suggests future enhancements.