A Comparison of the Performance of Artificial Neural Network Algorithms in Facial Expression Recognition

The methods for identifying facial expressions are presented in this research. The goal of this paper is to present a texture-oriented method combined with dimensional reduction that can be used to train the Single-Layer Neural Network (SLN), Back Propagation Algorithm (BPA), and Cerebellar Model Articulation Controller (CMAC) for identifying facial expressions. The proposed methods are known as intelligent methods because they can account for variations in facial expressions and thus perform better for untrained facial expressions. Conventional methods have limitations in that face expressions must adhere to certain guidelines. Gabor wavelet is used in different angles to extract possible textures of the facial expression to achieve expression detection accuracy. The higher dimensions of the extracted texture features are further reduced by using Fisher's linear discriminant function to improve the proposed method's accuracy. For training the proposed algorithms, Fisher's linear discriminant function is employed to turn a higher-dimensional feature vector into a two-dimensional vector. Angry, disgust, happiness, sadness, surprise, and fear are some of the facial emotions that are used. The proposed algorithms are compared in terms of performance.