Facial Expression Recognition using Weighted Mixture Deep Neural network Based on Double-channel Facial Images

Facial expression recognition has gained significant attention in recent years due to its applications in human-computer interaction, security, and social robotics. A promising approach involves using a Weighted Mixture Deep Neural Network (DNN) that leverages double-channel facial images. This method captures nuanced facial features by processing two channels of information simultaneously—often combining standard grayscale images with additional data, such as depth or infrared information. By assigning different weights to these channels, the network can prioritize the most relevant features for expression classification, improving accuracy. The architecture typically consists of convolutional layers that extract hierarchical features, followed by fully connected layers that make final predictions. This dual-channel approach enhances the model’s ability to generalize across diverse facial expressions and varying lighting conditions, ultimately leading to more robust and reliable recognition systems.