Enhanced Facial Micro Emotion Detection using Optimized Temporal Attention Recurrent Graph Convolutional Neural Network

Micro-expressions are brief natural facial expressions that reveal concealed emotions. However, traditional facial expression recognition (FER) methods struggled due to minute expression variations, limited dataset availability, and difference in lighting and camera angles, which complicates the classification process. Hence, this research proposes an Enhanced Facial Micro Emotion Detection using Optimized Temporal Attention Recurrent Graph Convolutional Neural Network (EFED-TARGCN), which effectively recognizes minute facial expressions. The framework integrates Low Dynamic Range Histogram Equalization (LDR-HE) for image enhancement, Revised Tunable Q-Factor Wavelet Transform (RTQWT) for spatial-temporal feature extraction, and Variable Velocity Strategy Particle Swarm Optimization (VVS-PSO) for relevant feature selection. Unlike conventional approaches that lack adaptive optimization, the proposed method introduces a Wolf-Bird Optimizer (WBO) to dynamically tune TARGCN parameters, enabling precise recognition of micro-expressions. The proposed method is implemented in python and evaluated using SMIC_preprocessed and SAMM v3 Datasets. The experimental results show that the proposed models attain higher accuracy of 99.47% on the SMIC_preprocessed dataset and 20.28%, 28.22% and 29.27% on the SAMM v3 Dataset compared with existing models. The proposed framework provides an effective solution for facial micro-expression recognition with potential applications in deception detection, psychological assessment, intelligent security systems and human-computer interaction.