Evaluating the Performance of Facial Expression Recognition Model via Various Classifiers

Facial expression recognition plays a major role in various application areas such as medical science, security, computer vision, etc. The face was detected and aligned using the ensemble of regression tree technique. The face detection and face alignment challenges such as pose variations, the camera distance from the face, and the lighting changes were analyzed. The aligned face region was connected and cropped with the boundary of the landmark points using Delaunay triangulation for extracting facial features. Two types of feature fusion methods are used to derive the distinctive aspect of the cropped face images. These features are fed into three classifiers to recognize six facial expressions. The error-correcting output code model achieves promising predictions in terms of accuracy.