Improving Facial Expression Recognition in HCI: Traditional Techniques and Comparative Analysis

Facial expressions play a vital role in human communication, enabling us to convey a wide range of emotions such as happiness, anger, and sadness. Human-Computer Interaction (HCI) is a rapidly growing and highly appealing field. By recognizing facial expressions, HCI can enhance communication by making it more interactive and natural. The proposed methodology is based on traditional techniques for extracting and classifying the features from images. To evaluate its effectiveness, we utilize two datasets: CK+ and JAFFE. First, the facial part of the images is cropped and enhanced using a CLAHE filter. Then, the Histogram of Oriented Gradients (HOG) descriptor extracts features from the images, while Principal Component Analysis (PCA) reduces their dimensionality. Finally, three classifiers Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Random Forest are used to classify facial expressions. Our methodology achieved an accuracy of 95.94% for CK+ and 93.75% for JAFFE datasets.