Automatic emotion detection according to facial expressions

Facial expressions are the most basic behaviors of people that are largely out of control and convey their feelings to the other person. Facial expressions play an important role in the interaction between people. It's the primary way people pass on the information they want to tell. Automatic recognition of facial expressions by computer is used in many different areas such as taking precautions against malicious people, diagnosing pain in newborn babies, detecting fatigue for safe driving, creating music playlists according to mood, and creating live emoji. interest. In this study, the determination of 7 facial expressions such as anger, disdain, disgust, fear, happiness, sadness and surprise on the images taken in real time by using artificial neural networks were examined. To determine the facial expressions, the triangulation points in the face region, which mark the exact locations of the eyebrows, eyes, nose, mouth and chin were used. Using these points, inferences were made about the status of facial expression markers such as mouth, eyebrow and eye in the face area. Distances between points are used for the status of each determinant or for relationships between determinants. With these distances, it is provided to express the conditions such as opening of the mouth, narrowing of the eye and lifting of the eyebrows. As a result of a comprehensive study, Cohn Kanade Extended image database has been used to reach 65% accuracy with artificial neural network trained on more than 4000 images. In this study, facial expressions can be detected from the images taken in real time regardless of the age, sex and skin color of the person with high accuracy rates, especially happy, confused and sad.