DEEP LEARNING APPROACH FOR FACIAL EXPRESSION RECOGNITION

Abstract

Facial expression refers to variations of faces in response to variations in human’s internal emotions. Face expres- sion classification and feature extraction are the two most impor- tant task performed in facial expression recognition. Converting facial expressions into numerical values using feature extraction techniques. There are various feature extraction algorithms avail- able. Principal component analysis, Linear Discriminant analysis are some of the facial feature extractors. Classifications are used for generating facial categories. Deep Learning based facial expression recognition algorithm becomes popular nowadays. Artificial Intelligence approach provides more accuracy and efficiency than old recognition methodologies. Social networking applications like Facebook, instagram uses newly developed deep learning algorithms for improved user experience. Proposed work consists of deep learning based facial expression recognition. Deep learning uses advanced machine level models that are backed with newest algorithms. A convolutional neural network is newer technique for classification and feature extraction. A deep learning open source python libraries called keras and tensorflow are used for implementing the facial expression recognition. The performance of the deep learning technique is measured using confusion matrix. Jaffe database, a Japanese person face images with various face emotions is used to test and train the deep learning model. The performance of the proposed technique is efficient and less computational.