Facial Expression Recognition Using Deep Learning Techniques

Our facial expressions play an important role in everyday communication between people. This automatic detection of facial expressions has long been studied for potential applications in various fields such as service robots, driver fatigue monitoring, and intelligent training systems. With the advent of human-computer interaction (HCI) systems such as social robots, visual interactive games, and data-driven animations, facial expression recognition (FER) has become a popular research area in recent years. Facial expressions convey 55% of messages conveyed, more than those conveyed by a combination of voice and language [5]. Face, voice, EEG and even text can be used to perform emotion recognition. Of these properties, facial expression is one of the most important for a number of reasons. It contains many useful features for recognizing emotions, it stands out, and it is easy to collect a large set of face data compared to other emotion recognition features. Facial expressions can be divided into six categories: anger, disgust, fear, surprise, sadness, and happiness. More recently, deep learning, especially Convolutional Neural Networks (CNNs), can be used to extract and train many features for proper face recognition systems. However, most of the clues come from different parts of the face, such as the mouth, nose, and eyes, while other parts, such as the hair, ears, and forehead, play a small role in the output. This means that, ideally, the machine learning system should focus only on the important parts of the face and be less sensitive to other areas of the face. In this paper, we propose a deep learning-based facial expression recognition framework that considers only important facial features and ignores other non-critical areas of the face.