A Comprehensive Development of Human Emotion Detection based on Facial Features using Digital Image Processing Methodology

This research introduces the Human Emotion Detection utilizing Facial Features (HEDFF) framework, presenting a comprehensive and advanced approach to emotion analysis. Focused on the integration of the AffectNet dataset, renowned for its diversity in annotated facial expressions, our methodology employs digital image processing techniques. Gabor filtering enhances relevant features crucial for subsequent analysis, followed by Convolutional Neural Network (CNN) utilization for robust feature extraction. The EfficientNet architecture refines emotion classification, optimizing efficiency. The AffectNet dataset serves as a cornerstone for model training and evaluation, contributing to the framework's efficacy. Gabor filtering during data preprocessing emphasizes critical textures. The CNN architecture captures intricate facial features, and EfficientNet optimizes classification efficiency. Particle Swarm Optimization (PSO) fine-tunes hyperparameters, enhancing the overall model performance. Rigorous evaluations against existing benchmarks showcase promising results in accuracy, precision, recall, and F1-score, with an outstanding overall accuracy of 98.87%. This research makes a significant contribution to the evolving landscape of emotion detection, providing a robust and efficient framework applicable across diverse domains. From human-computer interaction to mental health assessments, the proposed methodology demonstrates its versatility.