Expert System for Smart Virtual Facial Emotion Detection Using Convolutional Neural Network

Detecting facial emotions among people is a crucial task in social communication, as it reflects their internal character. In the future, virtual face emotion detection will play a vital role in various fields, such as virtual human detection, security systems, online games, human psychology analysis, virtual classrooms, and monitoring abnormalities in patients. Integrating facial emotion detection into virtual human detection enhances the entire virtual experience, infusing interactions with authenticity, emotional intelligence, and customization for individual users. Human emotions, depicted on the face represent the brain's reactions that can be captured in the form of video or image for accurate diagnosis. This paper introduces a technology-aided face emotion detection system using convolutional neural networks (CNN). The CNN model performs the emotion detection function by executing image pre-processing, feature extraction, and image classification. Computational modules within the neural network extract features from images to enhance prediction. The proposed CNN model uses data augmentation, max pooling, and batch normalization techniques to expand facial emotion classification and improve performance and generalization. Additionally, ResNet50 architecture used with CNN improves accuracy and reduces error rate with identity mapping. Comparing performance metrics, accuracy, loss and complexity to existing models, the proposed models outperform them. The proposed CNN achieves a maximum of 15.53% higher accuracy and 25.22% lower loss in face emotion detection than the lowest-performing existing model.