Enhanced CNN Architecture with Comprehensive Performance Metrics for Emotion Recognition

Engaging gameplay in video games provides a wide array of advantages, encompassing entertainment and personal growth. Immersive gaming experiences often stem from games that adapt to the player’s emotional state. In the proposed study, participants engage in games designed to elicit emotions like excitement, normalcy, and boredom. Their facial expressions and heart rate (HR) signals are monitored during gameplay. Facial expressions serve as ground truth indicators of the players’ emotions. These emotions, in tandem with the corresponding HR signals, are then processed to derive normalized features. These normalized features and identified emotions serve as input and output, respectively, for a Convolutional Neural Network (CNN) in a machine learning context. The trained CNN model subsequently enables the classification of participants’ emotions during gameplay. The experimental findings will not only identify the emotions but also gauge their intensity based on HR values, facilitating dynamic adjustments to the gameplay experience.