Abstract: The ability to analyse facial expressions plays a major role in non-verbal communication. If a someone only analyses what a person's mouth says and ignores what the person's face says, then we can only have a part of the story. Humans were the only ones who could distinguish between expressions but not anymore, with advancing technology our computers can learn how to detect emotions as well. This report is a guide to facial expression recognition software using OpenCV, Keras, Tensorflow and CNN, by implementing a program in Python it has become possible to build an algorithm that performs detection, extraction, and evaluation of these facial expressions for automatic recognition of human emotion in real-time. The main features of the face are considered for the detection of facial expressions. To determine the different emotions, the variations in each of the main features are used. To detect and classify different classes of emotions, machine learning algorithms are used by training different sets of images. This paper discusses a real-time emotion classification of a facial expression into one among the seven universal human expressions: Anger, Disgust, Fear, Happy, Neutral, Sad, Surprise by the implementation of a real-time vision system that can classify emotions.

Keywords: CNN, Facial Emotion Recognition, FACS, EMA.

PDF | DOI: 10.17148/IJARCCE.2021.105167

Open chat
Chat with IJARCCE