A Survey on Human Face Emotion Recognition using Machine Learning Models

Abstract

Human Face Emotion Recognition (HFER) is a field that studies the human state of mind through different facial expressions. In this paper, the comparative study is made on four different machine learning methods such as CNN, Bidirectional Convolutional LSTM, Multiple Pipelines, and Transfer Learning algorithms based on the various universal human emotions like Happy, Sad, Surprise, Fear, Anger, Disgust, Neutral and Contempt. Two Image data sets that carry static images with a different set of emotions are used in the methods and the accuracy level of each method is compared and analysed along with other metrics such as Precision, Recall, and F1 Scores are also considered.