Classification of Posed Smiles and Spontaneous Smiles with LSTM

Abstract

In recent years, the importance of smiles in communication has led to widespread research dealing with smiles. Attention has also been focused on discriminating posed and spontaneous facial expressions, suggesting that there is a difference in the movement of facial parts between posed and spontaneous smiles. In this paper, we focus on the fact that the time-series changes in facial expressions are important in the classification of smiles. We propose a method to classify posed smiles and spontaneous smiles by LSTM (Long short-term memory), which is often used for learning time series data. The classification result was 87.0%, indicating that it is useful to use LSTM to classify smiles.