Abstract

Facial expression recognition (FER) plays a critical role in human- computer interaction, enabling systems to interpret and respond to human emotions effectively. Traditional FER methods struggle to handle variations in viewpoint, illumination, and facial occlusions, leading to reduced accuracy and robustness. A neural network-based ensemble approach is proposed to address these challenges by combining the strengths of multiple deep learning models to enhance multi-view facial expression recognition. The proposed method integrates Convolutional Neural Networks (CNNs) with a stacking ensemble mechanism, where individual CNN models are trained on different facial angles and lighting conditions. The outputs of these models are combined using a meta-classifier to improve recognition accuracy. The ensemble network is trained and tested on the Multi-PIE and BU-3DFE datasets, achieving an accuracy of 96.4%, outperforming existing single-model approaches. The proposed method demonstrates robustness across varying facial poses and occlusions, highlighting its potential for real-world applications in emotion-aware systems and interactive technologies.