FER-Based Emotion-Informed Gender Classification

This paper introduces an innovative method for gender classification using facial images, integrating a Facial Emotion Recognition (FER) model to achieve superior accuracy. The proposed technique, termed Emotion-Informed Gender Classification (EIGC), leverages emotion-related features extracted from FER models to enhance the gender classification process. Unlike conventional deep learning models, which rely solely on facial geometry or features, EIGC incorporates emotional cues to improve classification performance. The method was rigorously tested on benchmark facial datasets, specifically the UTKFace and Kaggle gender classification datasets, where it achieved impressive accuracy rates of 97.64% and 97.54%, respectively. These results demonstrate a significant improvement, outper-forming widely used deep learning models such as CNN, AlexNet, VGG, and ResNet. The experimental evaluation underscores the robustness and effectiveness of the EIGC approach, highlighting the role of emotion-aware algorithms in enhancing gender classification accuracy. Overall, this work pushes the frontier of facial analysis and computer vision by illustrating the added value of integrating emotional features in gender classification tasks. This work opens new possibilities for emotion-informed models to be applied in other related domains of facial recognition and analysis.