REAL TIME FACIAL EMOTION RECOGNITION WITH DEEP CONVOLUTIONAL NEURAL NETWORK

Humans have the ability to recognize emotions through centuries, now we are making machines do the same. Emotion recognition can be done through body language, voice intonation, expressions. Though facial emotion recognition remains the most practical method. Facial expressions are triggered for a period of time as a response to the internal emotional state of a person. These also display social communications. Various attempts have been made which resulted in overcoming limitations and bringing new opportunities and to better understand and apply this simple way of human interaction in our world of computers. There has been the usage of new technologies for capturing facial expressions, with rapid, high-resolution image acquisition, these help us to analyze and recognize in real-time the true facial emotions. The present FER (Facial Expression Recognition) system uses still images, this faces a complex problem in discriminating foreground from background cluster without motion information. The FER in video motion is implemented to overcome this problem of existing systems. This paper brings an approach of Real-time Facial Expression Recognition using HAAR cascading classification for face detection followed by Convolutional Neural Networks (CNN) for classification of expressions. This model uses web-cam of the system and dynamically display the emotion in a text format. With an accuracy of 58% on test data, we were successfully able to classify seven different human emotions: Angry, Disgust, Fear, Happy, Sad, Surprise, Neutral. Facial emotion recognition can be used in many real-time applications, like airport security, trading, patient monitoring, and others.