An Optimized Convolution Neural Network Framework for Facial Expression Recognition

Abstract

Facial expression recognition (FER) is the simplest way to identify human emotions. Many machine and deep learning methods have been proposed to recognize human emotions from facial expressions. However, conventional machine learning methods suffer from poor feature representation and thus limited in performance. Therefore, deep learning methods have been preferred over them to represent features at micro level. Recently, convolution neural network (CNN) based deep models have gained popularity and widely explored for FER. However, hyperparameters tuning and overfitting avoidance is still challenging. Therefore, in this work, we propose convolution neural network based optimized FER to reduce overfitting by tuning the optimizer using data augmentation. We conducted several experiments to achieve better accuracy and used Adam optimizer for the proposed model. We have performed experiments over JAFFE and CK+ datasets and comparative analysis clearly validate the effectiveness of the proposed method in terms of accuracy, precision, recall and F1-score parameters.